Mercurial > repos > bgruening > sklearn_train_test_eval

annotate keras_train_and_eval.py @ 13:10e2498ef07b draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 8850f42c2c3763e614f7454c9c006f3d2ff572c0
author bgruening
date Fri, 27 May 2022 12:13:23 +0000
parents caf7d2b71a48
children 2eb5c017958d
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
1 import argparse
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
2 import json
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
3 import os
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
4 import pickle
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
5 import warnings
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
6 from itertools import chain
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
7
10
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
8 import joblib
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
9 import numpy as np
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
10 import pandas as pd
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
11 from galaxy_ml.externals.selene_sdk.utils import compute_score
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
12 from galaxy_ml.keras_galaxy_models import _predict_generator
10
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
13 from galaxy_ml.model_validations import train_test_split
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
14 from galaxy_ml.utils import (clean_params, get_main_estimator,
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
15 get_module, get_scoring, load_model, read_columns,
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
16 SafeEval, try_get_attr)
10
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
17 from scipy.io import mmread
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
18 from sklearn.metrics.scorer import _check_multimetric_scoring
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
19 from sklearn.model_selection import _search, _validation
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
20 from sklearn.model_selection._validation import _score
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
21 from sklearn.pipeline import Pipeline
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
22 from sklearn.utils import indexable, safe_indexing
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
23
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
24 _fit_and_score = try_get_attr("galaxy_ml.model_validations", "_fit_and_score")
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
25 setattr(_search, "_fit_and_score", _fit_and_score)
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
26 setattr(_validation, "_fit_and_score", _fit_and_score)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
27
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
28 N_JOBS = int(os.environ.get("GALAXY_SLOTS", 1))
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
29 CACHE_DIR = os.path.join(os.getcwd(), "cached")
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
30 del os
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
31 NON_SEARCHABLE = ("n_jobs", "pre_dispatch", "memory", "_path", "nthread", "callbacks")
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
32 ALLOWED_CALLBACKS = (
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
33 "EarlyStopping",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
34 "TerminateOnNaN",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
35 "ReduceLROnPlateau",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
36 "CSVLogger",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
37 "None",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
38 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
39
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
40
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
41 def _eval_swap_params(params_builder):
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
42 swap_params = {}
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
43
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
44 for p in params_builder["param_set"]:
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
45 swap_value = p["sp_value"].strip()
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
46 if swap_value == "":
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
47 continue
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
48
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
49 param_name = p["sp_name"]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
50 if param_name.lower().endswith(NON_SEARCHABLE):
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
51 warnings.warn(
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
52 "Warning: `%s` is not eligible for search and was "
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
53 "omitted!" % param_name
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
54 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
55 continue
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
56
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
57 if not swap_value.startswith(":"):
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
58 safe_eval = SafeEval(load_scipy=True, load_numpy=True)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
59 ev = safe_eval(swap_value)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
60 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
61 # Have `:` before search list, asks for estimator evaluatio
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
62 safe_eval_es = SafeEval(load_estimators=True)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
63 swap_value = swap_value[1:].strip()
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
64 # TODO maybe add regular express check
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
65 ev = safe_eval_es(swap_value)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
66
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
67 swap_params[param_name] = ev
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
68
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
69 return swap_params
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
70
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
71
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
72 def train_test_split_none(*arrays, **kwargs):
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
73 """extend train_test_split to take None arrays
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
74 and support split by group names.
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
75 """
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
76 nones = []
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
77 new_arrays = []
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
78 for idx, arr in enumerate(arrays):
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
79 if arr is None:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
80 nones.append(idx)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
81 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
82 new_arrays.append(arr)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
83
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
84 if kwargs["shuffle"] == "None":
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
85 kwargs["shuffle"] = None
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
86
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
87 group_names = kwargs.pop("group_names", None)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
88
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
89 if group_names is not None and group_names.strip():
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
90 group_names = [name.strip() for name in group_names.split(",")]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
91 new_arrays = indexable(*new_arrays)
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
92 groups = kwargs["labels"]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
93 n_samples = new_arrays[0].shape[0]
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
94 index_arr = np.arange(n_samples)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
95 test = index_arr[np.isin(groups, group_names)]
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
96 train = index_arr[~np.isin(groups, group_names)]
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
97 rval = list(
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
98 chain.from_iterable(
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
99 (safe_indexing(a, train), safe_indexing(a, test)) for a in new_arrays
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
100 )
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
101 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
102 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
103 rval = train_test_split(*new_arrays, **kwargs)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
104
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
105 for pos in nones:
10
a9e0b963b7bb "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 9
diff changeset
106 rval[pos * 2: 2] = [None, None]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
107
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
108 return rval
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
109
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
110
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
111 def _evaluate(y_true, pred_probas, scorer, is_multimetric=True):
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
112 """output scores based on input scorer
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
113
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
114 Parameters
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
115 ----------
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
116 y_true : array
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
117 True label or target values
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
118 pred_probas : array
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
119 Prediction values, probability for classification problem
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
120 scorer : dict
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
121 dict of `sklearn.metrics.scorer.SCORER`
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
122 is_multimetric : bool, default is True
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
123 """
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
124 if y_true.ndim == 1 or y_true.shape[-1] == 1:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
125 pred_probas = pred_probas.ravel()
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
126 pred_labels = (pred_probas > 0.5).astype("int32")
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
127 targets = y_true.ravel().astype("int32")
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
128 if not is_multimetric:
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
129 preds = (
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
130 pred_labels
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
131 if scorer.__class__.__name__ == "_PredictScorer"
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
132 else pred_probas
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
133 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
134 score = scorer._score_func(targets, preds, **scorer._kwargs)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
135
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
136 return score
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
137 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
138 scores = {}
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
139 for name, one_scorer in scorer.items():
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
140 preds = (
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
141 pred_labels
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
142 if one_scorer.__class__.__name__ == "_PredictScorer"
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
143 else pred_probas
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
144 )
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
145 score = one_scorer._score_func(targets, preds, **one_scorer._kwargs)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
146 scores[name] = score
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
147
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
148 # TODO: multi-class metrics
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
149 # multi-label
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
150 else:
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
151 pred_labels = (pred_probas > 0.5).astype("int32")
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
152 targets = y_true.astype("int32")
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
153 if not is_multimetric:
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
154 preds = (
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
155 pred_labels
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
156 if scorer.__class__.__name__ == "_PredictScorer"
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
157 else pred_probas
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
158 )
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
159 score, _ = compute_score(preds, targets, scorer._score_func)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
160 return score
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
161 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
162 scores = {}
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
163 for name, one_scorer in scorer.items():
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
164 preds = (
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
165 pred_labels
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
166 if one_scorer.__class__.__name__ == "_PredictScorer"
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
167 else pred_probas
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
168 )
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
169 score, _ = compute_score(preds, targets, one_scorer._score_func)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
170 scores[name] = score
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
171
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
172 return scores
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
173
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
174
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
175 def main(
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
176 inputs,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
177 infile_estimator,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
178 infile1,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
179 infile2,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
180 outfile_result,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
181 outfile_object=None,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
182 outfile_weights=None,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
183 outfile_y_true=None,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
184 outfile_y_preds=None,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
185 groups=None,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
186 ref_seq=None,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
187 intervals=None,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
188 targets=None,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
189 fasta_path=None,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
190 ):
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
191 """
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
192 Parameter
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
193 ---------
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
194 inputs : str
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
195 File path to galaxy tool parameter
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
196
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
197 infile_estimator : str
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
198 File path to estimator
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
199
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
200 infile1 : str
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
201 File path to dataset containing features
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
202
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
203 infile2 : str
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
204 File path to dataset containing target values
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
205
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
206 outfile_result : str
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
207 File path to save the results, either cv_results or test result
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
208
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
209 outfile_object : str, optional
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
210 File path to save searchCV object
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
211
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
212 outfile_weights : str, optional
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
213 File path to save deep learning model weights
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
214
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
215 outfile_y_true : str, optional
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
216 File path to target values for prediction
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
217
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
218 outfile_y_preds : str, optional
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
219 File path to save deep learning model weights
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
220
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
221 groups : str
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
222 File path to dataset containing groups labels
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
223
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
224 ref_seq : str
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
225 File path to dataset containing genome sequence file
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
226
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
227 intervals : str
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
228 File path to dataset containing interval file
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
229
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
230 targets : str
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
231 File path to dataset compressed target bed file
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
232
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
233 fasta_path : str
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
234 File path to dataset containing fasta file
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
235 """
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
236 warnings.simplefilter("ignore")
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
237
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
238 with open(inputs, "r") as param_handler:
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
239 params = json.load(param_handler)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
240
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
241 # load estimator
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
242 with open(infile_estimator, "rb") as estimator_handler:
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
243 estimator = load_model(estimator_handler)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
244
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
245 estimator = clean_params(estimator)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
246
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
247 # swap hyperparameter
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
248 swapping = params["experiment_schemes"]["hyperparams_swapping"]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
249 swap_params = _eval_swap_params(swapping)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
250 estimator.set_params(**swap_params)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
251
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
252 estimator_params = estimator.get_params()
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
253
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
254 # store read dataframe object
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
255 loaded_df = {}
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
256
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
257 input_type = params["input_options"]["selected_input"]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
258 # tabular input
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
259 if input_type == "tabular":
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
260 header = "infer" if params["input_options"]["header1"] else None
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
261 column_option = params["input_options"]["column_selector_options_1"][
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
262 "selected_column_selector_option"
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
263 ]
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
264 if column_option in [
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
265 "by_index_number",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
266 "all_but_by_index_number",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
267 "by_header_name",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
268 "all_but_by_header_name",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
269 ]:
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
270 c = params["input_options"]["column_selector_options_1"]["col1"]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
271 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
272 c = None
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
273
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
274 df_key = infile1 + repr(header)
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
275 df = pd.read_csv(infile1, sep="\t", header=header, parse_dates=True)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
276 loaded_df[df_key] = df
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
277
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
278 X = read_columns(df, c=c, c_option=column_option).astype(float)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
279 # sparse input
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
280 elif input_type == "sparse":
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
281 X = mmread(open(infile1, "r"))
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
282
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
283 # fasta_file input
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
284 elif input_type == "seq_fasta":
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
285 pyfaidx = get_module("pyfaidx")
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
286 sequences = pyfaidx.Fasta(fasta_path)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
287 n_seqs = len(sequences.keys())
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
288 X = np.arange(n_seqs)[:, np.newaxis]
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
289 for param in estimator_params.keys():
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
290 if param.endswith("fasta_path"):
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
291 estimator.set_params(**{param: fasta_path})
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
292 break
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
293 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
294 raise ValueError(
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
295 "The selected estimator doesn't support "
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
296 "fasta file input! Please consider using "
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
297 "KerasGBatchClassifier with "
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
298 "FastaDNABatchGenerator/FastaProteinBatchGenerator "
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
299 "or having GenomeOneHotEncoder/ProteinOneHotEncoder "
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
300 "in pipeline!"
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
301 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
302
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
303 elif input_type == "refseq_and_interval":
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
304 path_params = {
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
305 "data_batch_generator__ref_genome_path": ref_seq,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
306 "data_batch_generator__intervals_path": intervals,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
307 "data_batch_generator__target_path": targets,
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
308 }
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
309 estimator.set_params(**path_params)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
310 n_intervals = sum(1 for line in open(intervals))
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
311 X = np.arange(n_intervals)[:, np.newaxis]
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
312
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
313 # Get target y
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
314 header = "infer" if params["input_options"]["header2"] else None
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
315 column_option = params["input_options"]["column_selector_options_2"][
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
316 "selected_column_selector_option2"
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
317 ]
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
318 if column_option in [
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
319 "by_index_number",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
320 "all_but_by_index_number",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
321 "by_header_name",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
322 "all_but_by_header_name",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
323 ]:
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
324 c = params["input_options"]["column_selector_options_2"]["col2"]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
325 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
326 c = None
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
327
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
328 df_key = infile2 + repr(header)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
329 if df_key in loaded_df:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
330 infile2 = loaded_df[df_key]
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
331 else:
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
332 infile2 = pd.read_csv(infile2, sep="\t", header=header, parse_dates=True)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
333 loaded_df[df_key] = infile2
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
334
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
335 y = read_columns(
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
336 infile2, c=c, c_option=column_option, sep="\t", header=header, parse_dates=True
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
337 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
338 if len(y.shape) == 2 and y.shape[1] == 1:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
339 y = y.ravel()
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
340 if input_type == "refseq_and_interval":
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
341 estimator.set_params(data_batch_generator__features=y.ravel().tolist())
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
342 y = None
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
343 # end y
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
344
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
345 # load groups
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
346 if groups:
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
347 groups_selector = (
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
348 params["experiment_schemes"]["test_split"]["split_algos"]
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
349 ).pop("groups_selector")
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
350
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
351 header = "infer" if groups_selector["header_g"] else None
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
352 column_option = groups_selector["column_selector_options_g"][
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
353 "selected_column_selector_option_g"
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
354 ]
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
355 if column_option in [
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
356 "by_index_number",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
357 "all_but_by_index_number",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
358 "by_header_name",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
359 "all_but_by_header_name",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
360 ]:
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
361 c = groups_selector["column_selector_options_g"]["col_g"]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
362 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
363 c = None
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
364
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
365 df_key = groups + repr(header)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
366 if df_key in loaded_df:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
367 groups = loaded_df[df_key]
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
368
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
369 groups = read_columns(
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
370 groups,
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
371 c=c,
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
372 c_option=column_option,
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
373 sep="\t",
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
374 header=header,
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
375 parse_dates=True,
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
376 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
377 groups = groups.ravel()
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
378
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
379 # del loaded_df
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
380 del loaded_df
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
381
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
382 # cache iraps_core fits could increase search speed significantly
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
383 memory = joblib.Memory(location=CACHE_DIR, verbose=0)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
384 main_est = get_main_estimator(estimator)
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
385 if main_est.__class__.__name__ == "IRAPSClassifier":
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
386 main_est.set_params(memory=memory)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
387
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
388 # handle scorer, convert to scorer dict
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
389 scoring = params["experiment_schemes"]["metrics"]["scoring"]
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
390 if scoring is not None:
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
391 # get_scoring() expects secondary_scoring to be a comma separated string (not a list)
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
392 # Check if secondary_scoring is specified
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
393 secondary_scoring = scoring.get("secondary_scoring", None)
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
394 if secondary_scoring is not None:
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
395 # If secondary_scoring is specified, convert the list into comman separated string
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
396 scoring["secondary_scoring"] = ",".join(scoring["secondary_scoring"])
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
397
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
398 scorer = get_scoring(scoring)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
399 scorer, _ = _check_multimetric_scoring(estimator, scoring=scorer)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
400
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
401 # handle test (first) split
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
402 test_split_options = params["experiment_schemes"]["test_split"]["split_algos"]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
403
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
404 if test_split_options["shuffle"] == "group":
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
405 test_split_options["labels"] = groups
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
406 if test_split_options["shuffle"] == "stratified":
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
407 if y is not None:
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
408 test_split_options["labels"] = y
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
409 else:
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
410 raise ValueError(
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
411 "Stratified shuffle split is not " "applicable on empty target values!"
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
412 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
413
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
414 (
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
415 X_train,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
416 X_test,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
417 y_train,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
418 y_test,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
419 groups_train,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
420 _groups_test,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
421 ) = train_test_split_none(X, y, groups, **test_split_options)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
422
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
423 exp_scheme = params["experiment_schemes"]["selected_exp_scheme"]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
424
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
425 # handle validation (second) split
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
426 if exp_scheme == "train_val_test":
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
427 val_split_options = params["experiment_schemes"]["val_split"]["split_algos"]
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
428
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
429 if val_split_options["shuffle"] == "group":
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
430 val_split_options["labels"] = groups_train
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
431 if val_split_options["shuffle"] == "stratified":
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
432 if y_train is not None:
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
433 val_split_options["labels"] = y_train
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
434 else:
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
435 raise ValueError(
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
436 "Stratified shuffle split is not "
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
437 "applicable on empty target values!"
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
438 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
439
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
440 (
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
441 X_train,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
442 X_val,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
443 y_train,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
444 y_val,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
445 groups_train,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
446 _groups_val,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
447 ) = train_test_split_none(X_train, y_train, groups_train, **val_split_options)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
448
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
449 # train and eval
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
450 if hasattr(estimator, "validation_data"):
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
451 if exp_scheme == "train_val_test":
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
452 estimator.fit(X_train, y_train, validation_data=(X_val, y_val))
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
453 else:
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
454 estimator.fit(X_train, y_train, validation_data=(X_test, y_test))
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
455 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
456 estimator.fit(X_train, y_train)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
457
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
458 if hasattr(estimator, "evaluate"):
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
459 steps = estimator.prediction_steps
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
460 batch_size = estimator.batch_size
11
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
461 generator = estimator.data_generator_.flow(
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
462 X_test, y=y_test, batch_size=batch_size
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
463 )
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
464 predictions, y_true = _predict_generator(
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
465 estimator.model_, generator, steps=steps
caf7d2b71a48 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 10
diff changeset
466 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
467 scores = _evaluate(y_true, predictions, scorer, is_multimetric=True)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
468
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
469 else:
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
470 if hasattr(estimator, "predict_proba"):
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
471 predictions = estimator.predict_proba(X_test)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
472 else:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
473 predictions = estimator.predict(X_test)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
474
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
475 y_true = y_test
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
476 scores = _score(estimator, X_test, y_test, scorer, is_multimetric=True)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
477 if outfile_y_true:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
478 try:
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
479 pd.DataFrame(y_true).to_csv(outfile_y_true, sep="\t", index=False)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
480 pd.DataFrame(predictions).astype(np.float32).to_csv(
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
481 outfile_y_preds,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
482 sep="\t",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
483 index=False,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
484 float_format="%g",
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
485 chunksize=10000,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
486 )
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
487 except Exception as e:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
488 print("Error in saving predictions: %s" % e)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
489
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
490 # handle output
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
491 for name, score in scores.items():
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
492 scores[name] = [score]
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
493 df = pd.DataFrame(scores)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
494 df = df[sorted(df.columns)]
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
495 df.to_csv(path_or_buf=outfile_result, sep="\t", header=True, index=False)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
496
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
497 memory.clear(warn=False)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
498
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
499 if outfile_object:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
500 main_est = estimator
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
501 if isinstance(estimator, Pipeline):
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
502 main_est = estimator.steps[-1][-1]
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
503
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
504 if hasattr(main_est, "model_") and hasattr(main_est, "save_weights"):
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
505 if outfile_weights:
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
506 main_est.save_weights(outfile_weights)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
507 del main_est.model_
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
508 del main_est.fit_params
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
509 del main_est.model_class_
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
510 if getattr(main_est, "validation_data", None):
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
511 del main_est.validation_data
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
512 if getattr(main_est, "data_generator_", None):
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
513 del main_est.data_generator_
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
514
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
515 with open(outfile_object, "wb") as output_handler:
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
516 pickle.dump(estimator, output_handler, pickle.HIGHEST_PROTOCOL)
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
517
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
518
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
519 if __name__ == "__main__":
5
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
520 aparser = argparse.ArgumentParser()
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
521 aparser.add_argument("-i", "--inputs", dest="inputs", required=True)
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
522 aparser.add_argument("-e", "--estimator", dest="infile_estimator")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
523 aparser.add_argument("-X", "--infile1", dest="infile1")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
524 aparser.add_argument("-y", "--infile2", dest="infile2")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
525 aparser.add_argument("-O", "--outfile_result", dest="outfile_result")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
526 aparser.add_argument("-o", "--outfile_object", dest="outfile_object")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
527 aparser.add_argument("-w", "--outfile_weights", dest="outfile_weights")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
528 aparser.add_argument("-l", "--outfile_y_true", dest="outfile_y_true")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
529 aparser.add_argument("-p", "--outfile_y_preds", dest="outfile_y_preds")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
530 aparser.add_argument("-g", "--groups", dest="groups")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
531 aparser.add_argument("-r", "--ref_seq", dest="ref_seq")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
532 aparser.add_argument("-b", "--intervals", dest="intervals")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
533 aparser.add_argument("-t", "--targets", dest="targets")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
534 aparser.add_argument("-f", "--fasta_path", dest="fasta_path")
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
535 args = aparser.parse_args()
2b8406e74f9e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
536
9
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
537 main(
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
538 args.inputs,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
539 args.infile_estimator,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
540 args.infile1,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
541 args.infile2,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
542 args.outfile_result,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
543 outfile_object=args.outfile_object,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
544 outfile_weights=args.outfile_weights,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
545 outfile_y_true=args.outfile_y_true,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
546 outfile_y_preds=args.outfile_y_preds,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
547 groups=args.groups,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
548 ref_seq=args.ref_seq,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
549 intervals=args.intervals,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
550 targets=args.targets,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
551 fasta_path=args.fasta_path,
ead7adad8d0e "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
552 )