Mercurial > repos > goeckslab > ludwig_train

annotate ludwig_experiment.py @ 0:f0be10937f5c draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
author goeckslab
date Tue, 07 Jan 2025 22:44:09 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
1 import json
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
2 import logging
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
3 import os
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
4 import pickle
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
5 import sys
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
6
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
7 from jinja_report import generate_report
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
8
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
9 from ludwig.experiment import cli
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
10 from ludwig.globals import (
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
11 DESCRIPTION_FILE_NAME,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
12 PREDICTIONS_PARQUET_FILE_NAME,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
13 TEST_STATISTICS_FILE_NAME,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
14 TRAIN_SET_METADATA_FILE_NAME
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
15 )
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
16 from ludwig.utils.data_utils import get_split_path
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
17 from ludwig.visualize import get_visualizations_registry
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
18
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
19 from model_unpickler import SafeUnpickler
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
20
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
21 import pandas as pd
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
22
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
23 from utils import (
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
24 encode_image_to_base64,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
25 get_html_closing,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
26 get_html_template
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
27 )
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
28
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
29 import yaml
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
30
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
31
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
32 logging.basicConfig(level=logging.DEBUG)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
33
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
34 LOG = logging.getLogger(__name__)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
35
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
36 setattr(pickle, 'Unpickler', SafeUnpickler)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
37
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
38 # visualization
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
39 output_directory = None
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
40 for ix, arg in enumerate(sys.argv):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
41 if arg == "--output_directory":
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
42 output_directory = sys.argv[ix+1]
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
43 break
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
44
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
45 viz_output_directory = os.path.join(output_directory, "visualizations")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
46
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
47
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
48 def get_output_feature_name(experiment_dir, output_feature=0):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
49 """Helper function to extract specified output feature name.
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
50
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
51 :param experiment_dir: Path to the experiment directory
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
52 :param output_feature: position of the output feature the description.json
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
53 :return output_feature_name: name of the first output feature name
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
54 from the experiment
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
55 """
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
56 if os.path.exists(os.path.join(experiment_dir, DESCRIPTION_FILE_NAME)):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
57 description_file = os.path.join(experiment_dir, DESCRIPTION_FILE_NAME)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
58 with open(description_file, "rb") as f:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
59 content = json.load(f)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
60 output_feature_name = \
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
61 content["config"]["output_features"][output_feature]["name"]
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
62 dataset_path = content["dataset"]
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
63 return output_feature_name, dataset_path
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
64 return None, None
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
65
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
66
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
67 def check_file(file_path):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
68 """Check if the file exists; return None if it doesn't."""
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
69 return file_path if os.path.exists(file_path) else None
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
70
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
71
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
72 def make_visualizations(ludwig_output_directory_name):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
73 ludwig_output_directory = os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
74 output_directory,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
75 ludwig_output_directory_name,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
76 )
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
77 visualizations = [
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
78 "confidence_thresholding",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
79 "confidence_thresholding_data_vs_acc",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
80 "confidence_thresholding_data_vs_acc_subset",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
81 "confidence_thresholding_data_vs_acc_subset_per_class",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
82 "confidence_thresholding_2thresholds_2d",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
83 "confidence_thresholding_2thresholds_3d",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
84 "binary_threshold_vs_metric",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
85 "roc_curves",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
86 "roc_curves_from_test_statistics",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
87 "calibration_1_vs_all",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
88 "calibration_multiclass",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
89 "confusion_matrix",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
90 "frequency_vs_f1",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
91 "learning_curves",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
92 ]
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
93
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
94 # Check existence of required files
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
95 training_statistics = check_file(os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
96 ludwig_output_directory,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
97 "training_statistics.json",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
98 ))
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
99 test_statistics = check_file(os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
100 ludwig_output_directory,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
101 TEST_STATISTICS_FILE_NAME,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
102 ))
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
103 ground_truth_metadata = check_file(os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
104 ludwig_output_directory,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
105 "model",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
106 TRAIN_SET_METADATA_FILE_NAME,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
107 ))
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
108 probabilities = check_file(os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
109 ludwig_output_directory,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
110 PREDICTIONS_PARQUET_FILE_NAME,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
111 ))
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
112
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
113 output_feature, dataset_path = get_output_feature_name(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
114 ludwig_output_directory)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
115 ground_truth = None
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
116 split_file = None
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
117 if dataset_path:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
118 ground_truth = check_file(dataset_path)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
119 split_file = check_file(get_split_path(dataset_path))
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
120
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
121 if (not output_feature) and (test_statistics):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
122 test_stat = os.path.join(test_statistics)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
123 with open(test_stat, "rb") as f:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
124 content = json.load(f)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
125 output_feature = next(iter(content.keys()))
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
126
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
127 for viz in visualizations:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
128 viz_func = get_visualizations_registry()[viz]
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
129 try:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
130 viz_func(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
131 training_statistics=[training_statistics]
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
132 if training_statistics else [],
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
133 test_statistics=[test_statistics] if test_statistics else [],
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
134 probabilities=[probabilities] if probabilities else [],
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
135 top_n_classes=[0],
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
136 output_feature_name=output_feature if output_feature else "",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
137 ground_truth_split=2,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
138 top_k=3,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
139 ground_truth_metadata=ground_truth_metadata,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
140 ground_truth=ground_truth,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
141 split_file=split_file,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
142 output_directory=viz_output_directory,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
143 normalize=False,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
144 file_format="png",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
145 )
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
146 except Exception as e:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
147 LOG.info(f"Visualization: {viz}")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
148 LOG.info(f"Error: {e}")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
149
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
150
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
151 # report
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
152 def render_report(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
153 title: str,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
154 ludwig_output_directory_name: str,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
155 show_visualization: bool = True
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
156 ):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
157 ludwig_output_directory = os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
158 output_directory,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
159 ludwig_output_directory_name,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
160 )
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
161 report_config = {
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
162 "title": title,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
163 }
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
164 if show_visualization:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
165 report_config["visualizations"] = [
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
166 {
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
167 "src": f"visualizations/{fl}",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
168 "type": "image" if fl[fl.rindex(".") + 1:] == "png" else
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
169 fl[fl.rindex(".") + 1:],
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
170 } for fl in sorted(os.listdir(viz_output_directory))
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
171 ]
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
172 report_config["raw outputs"] = [
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
173 {
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
174 "src": f"{fl}",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
175 "type": "json" if fl.endswith(".json") else "unclassified",
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
176 } for fl in sorted(os.listdir(ludwig_output_directory))
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
177 if fl.endswith((".json", ".parquet"))
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
178 ]
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
179
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
180 with open(os.path.join(output_directory, "report_config.yml"), 'w') as fh:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
181 yaml.safe_dump(report_config, fh)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
182
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
183 report_path = os.path.join(output_directory, "smart_report.html")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
184 generate_report.main(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
185 report_config,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
186 schema={"html_height": 800},
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
187 outfile=report_path,
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
188 )
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
189
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
190
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
191 def convert_parquet_to_csv(ludwig_output_directory_name):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
192 """Convert the predictions Parquet file to CSV."""
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
193 ludwig_output_directory = os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
194 output_directory, ludwig_output_directory_name)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
195 parquet_path = os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
196 ludwig_output_directory, "predictions.parquet")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
197 csv_path = os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
198 ludwig_output_directory, "predictions_parquet.csv")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
199
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
200 try:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
201 df = pd.read_parquet(parquet_path)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
202 df.to_csv(csv_path, index=False)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
203 LOG.info(f"Converted Parquet to CSV: {csv_path}")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
204 except Exception as e:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
205 LOG.error(f"Error converting Parquet to CSV: {e}")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
206
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
207
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
208 def generate_html_report(title, ludwig_output_directory_name):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
209 # ludwig_output_directory = os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
210 # output_directory, ludwig_output_directory_name)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
211
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
212 # test_statistics_html = ""
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
213 # # Read test statistics JSON and convert to HTML table
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
214 # try:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
215 # test_statistics_path = os.path.join(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
216 # ludwig_output_directory, TEST_STATISTICS_FILE_NAME)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
217 # with open(test_statistics_path, "r") as f:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
218 # test_statistics = json.load(f)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
219 # test_statistics_html = "<h2>Test Statistics</h2>"
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
220 # test_statistics_html += json_to_html_table(
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
221 # test_statistics)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
222 # except Exception as e:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
223 # LOG.info(f"Error reading test statistics: {e}")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
224
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
225 # Convert visualizations to HTML
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
226 plots_html = ""
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
227 if len(os.listdir(viz_output_directory)) > 0:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
228 plots_html = "<h2>Visualizations</h2>"
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
229 for plot_file in sorted(os.listdir(viz_output_directory)):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
230 plot_path = os.path.join(viz_output_directory, plot_file)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
231 if os.path.isfile(plot_path) and plot_file.endswith((".png", ".jpg")):
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
232 encoded_image = encode_image_to_base64(plot_path)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
233 plots_html += (
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
234 f'<div class="plot">'
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
235 f'<h3>{os.path.splitext(plot_file)[0]}</h3>'
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
236 '<img src="data:image/png;base64,'
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
237 f'{encoded_image}" alt="{plot_file}">'
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
238 f'</div>'
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
239 )
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
240
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
241 # Generate the full HTML content
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
242 html_content = f"""
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
243 {get_html_template()}
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
244 <h1>{title}</h1>
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
245 {plots_html}
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
246 {get_html_closing()}
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
247 """
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
248
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
249 # Save the HTML report
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
250 title: str
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
251 report_name = title.lower().replace(" ", "_")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
252 report_path = os.path.join(output_directory, f"{report_name}_report.html")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
253 with open(report_path, "w") as report_file:
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
254 report_file.write(html_content)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
255
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
256 LOG.info(f"HTML report generated at: {report_path}")
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
257
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
258
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
259 if __name__ == "__main__":
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
260
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
261 cli(sys.argv[1:])
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
262
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
263 ludwig_output_directory_name = "experiment_run"
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
264
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
265 make_visualizations(ludwig_output_directory_name)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
266 # title = "Ludwig Experiment"
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
267 # render_report(title, ludwig_output_directory_name)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
268 convert_parquet_to_csv(ludwig_output_directory_name)
f0be10937f5c planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff changeset
269 generate_html_report("Ludwig Experiment", ludwig_output_directory_name)