Mercurial > repos > bgruening > sklearn_svm_classifier
annotate fitted_model_eval.py @ 18:02d2be90dedb draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
author | bgruening |
---|---|
date | Thu, 01 Oct 2020 20:06:56 +0000 |
parents | 9295c5f34630 |
children | d67dcd63f6cb |
rev | line source |
---|---|
13
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
1 import argparse |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
2 import json |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
3 import pandas as pd |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
4 import warnings |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
5 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
6 from scipy.io import mmread |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
7 from sklearn.pipeline import Pipeline |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
8 from sklearn.metrics.scorer import _check_multimetric_scoring |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
9 from sklearn.model_selection._validation import _score |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
10 from galaxy_ml.utils import get_scoring, load_model, read_columns |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
11 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
12 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
13 def _get_X_y(params, infile1, infile2): |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
14 """ read from inputs and output X and y |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
15 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
16 Parameters |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
17 ---------- |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
18 params : dict |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
19 Tool inputs parameter |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
20 infile1 : str |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
21 File path to dataset containing features |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
22 infile2 : str |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
23 File path to dataset containing target values |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
24 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
25 """ |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
26 # store read dataframe object |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
27 loaded_df = {} |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
28 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
29 input_type = params['input_options']['selected_input'] |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
30 # tabular input |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
31 if input_type == 'tabular': |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
32 header = 'infer' if params['input_options']['header1'] else None |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
33 column_option = (params['input_options']['column_selector_options_1'] |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
34 ['selected_column_selector_option']) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
35 if column_option in ['by_index_number', 'all_but_by_index_number', |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
36 'by_header_name', 'all_but_by_header_name']: |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
37 c = params['input_options']['column_selector_options_1']['col1'] |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
38 else: |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
39 c = None |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
40 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
41 df_key = infile1 + repr(header) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
42 df = pd.read_csv(infile1, sep='\t', header=header, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
43 parse_dates=True) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
44 loaded_df[df_key] = df |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
45 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
46 X = read_columns(df, c=c, c_option=column_option).astype(float) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
47 # sparse input |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
48 elif input_type == 'sparse': |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
49 X = mmread(open(infile1, 'r')) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
50 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
51 # Get target y |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
52 header = 'infer' if params['input_options']['header2'] else None |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
53 column_option = (params['input_options']['column_selector_options_2'] |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
54 ['selected_column_selector_option2']) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
55 if column_option in ['by_index_number', 'all_but_by_index_number', |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
56 'by_header_name', 'all_but_by_header_name']: |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
57 c = params['input_options']['column_selector_options_2']['col2'] |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
58 else: |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
59 c = None |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
60 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
61 df_key = infile2 + repr(header) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
62 if df_key in loaded_df: |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
63 infile2 = loaded_df[df_key] |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
64 else: |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
65 infile2 = pd.read_csv(infile2, sep='\t', |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
66 header=header, parse_dates=True) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
67 loaded_df[df_key] = infile2 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
68 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
69 y = read_columns( |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
70 infile2, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
71 c=c, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
72 c_option=column_option, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
73 sep='\t', |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
74 header=header, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
75 parse_dates=True) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
76 if len(y.shape) == 2 and y.shape[1] == 1: |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
77 y = y.ravel() |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
78 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
79 return X, y |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
80 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
81 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
82 def main(inputs, infile_estimator, outfile_eval, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
83 infile_weights=None, infile1=None, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
84 infile2=None): |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
85 """ |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
86 Parameter |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
87 --------- |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
88 inputs : str |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
89 File path to galaxy tool parameter |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
90 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
91 infile_estimator : strgit |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
92 File path to trained estimator input |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
93 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
94 outfile_eval : str |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
95 File path to save the evalulation results, tabular |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
96 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
97 infile_weights : str |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
98 File path to weights input |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
99 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
100 infile1 : str |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
101 File path to dataset containing features |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
102 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
103 infile2 : str |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
104 File path to dataset containing target values |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
105 """ |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
106 warnings.filterwarnings('ignore') |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
107 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
108 with open(inputs, 'r') as param_handler: |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
109 params = json.load(param_handler) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
110 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
111 X_test, y_test = _get_X_y(params, infile1, infile2) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
112 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
113 # load model |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
114 with open(infile_estimator, 'rb') as est_handler: |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
115 estimator = load_model(est_handler) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
116 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
117 main_est = estimator |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
118 if isinstance(estimator, Pipeline): |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
119 main_est = estimator.steps[-1][-1] |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
120 if hasattr(main_est, 'config') and hasattr(main_est, 'load_weights'): |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
121 if not infile_weights or infile_weights == 'None': |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
122 raise ValueError("The selected model skeleton asks for weights, " |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
123 "but no dataset for weights was provided!") |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
124 main_est.load_weights(infile_weights) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
125 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
126 # handle scorer, convert to scorer dict |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
127 scoring = params['scoring'] |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
128 scorer = get_scoring(scoring) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
129 scorer, _ = _check_multimetric_scoring(estimator, scoring=scorer) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
130 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
131 if hasattr(estimator, 'evaluate'): |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
132 scores = estimator.evaluate(X_test, y_test=y_test, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
133 scorer=scorer, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
134 is_multimetric=True) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
135 else: |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
136 scores = _score(estimator, X_test, y_test, scorer, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
137 is_multimetric=True) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
138 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
139 # handle output |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
140 for name, score in scores.items(): |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
141 scores[name] = [score] |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
142 df = pd.DataFrame(scores) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
143 df = df[sorted(df.columns)] |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
144 df.to_csv(path_or_buf=outfile_eval, sep='\t', |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
145 header=True, index=False) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
146 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
147 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
148 if __name__ == '__main__': |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
149 aparser = argparse.ArgumentParser() |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
150 aparser.add_argument("-i", "--inputs", dest="inputs", required=True) |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
151 aparser.add_argument("-e", "--infile_estimator", dest="infile_estimator") |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
152 aparser.add_argument("-w", "--infile_weights", dest="infile_weights") |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
153 aparser.add_argument("-X", "--infile1", dest="infile1") |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
154 aparser.add_argument("-y", "--infile2", dest="infile2") |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
155 aparser.add_argument("-O", "--outfile_eval", dest="outfile_eval") |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
156 args = aparser.parse_args() |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
157 |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
158 main(args.inputs, args.infile_estimator, args.outfile_eval, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
159 infile_weights=args.infile_weights, infile1=args.infile1, |
9295c5f34630
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit eb703290e2589561ea215c84aa9f71bcfe1712c6"
bgruening
parents:
diff
changeset
|
160 infile2=args.infile2) |