sk-factor/sk_factor.py at main · B2F/sk-factor · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
import argparse
import pandas as pd
import re
import sys

from src.engine.config import Config
from src.engine.preprocessors import Preprocessors
from src.engine.transfomers import Transformers
from src.engine.training import Training
from src.engine.plots import Plots
from src.engine.debug import Debugger
from src.engine.predictions import Predictions
from src.engine.plugins import Plugins
from src.engine.files import Files

def main():

    parser = argparse.ArgumentParser()

    parser.add_argument("-c", "--config", help = "Use a config file from the config/ directory", default='sk_factor')

    # -t and -p arguments can be cumulated
    parser.add_argument("-t", "--train_files", help = "Train with given file(s)", required = False, nargs = "*")
    parser.add_argument("-p", "--predict_files", help = "Predict with given file(s)", required = False, nargs = "*")
    parser.add_argument("-m", "--model_file", help = "Model file(s) used for predictions", required = False, nargs = "*")

    parser.add_argument("-d", "--debug", help = "Enable debugging", action='store_true')

    group = parser.add_mutually_exclusive_group()
    group.add_argument("-ef", "--explore", help = "EDA plots only", action='store_true', required = False)
    group.add_argument("-tf", "--train", help = "Training only", action='store_true', required = False)
    group.add_argument("-pf", "--predict", help = "Predict only", action='store_true', required = False)

    # Show help if no arguments provided
    if len(sys.argv) == 1:
        parser.print_help()
        return

    argument = parser.parse_args()

    config = Config(argument.config)
    reConfig = re.search(r"(?:.*/)?([^\/\.]*)(?:\.toml)$", argument.config)

    Debugger.attach(config)

    config.set('dataset', 'filename', reConfig.group(1))

    config.set('debug', 'enabled', True) if argument.debug else config.set('debug', 'enabled', False)

    if argument.explore:
        config.set('eda', 'enabled', True)
        config.set('training', 'enabled', False)
        config.set('predictions', 'enabled', False)
    elif argument.train:
        config.set('eda', 'enabled', False)
        config.set('training', 'enabled', True)
        config.set('predictions', 'enabled', False)
    elif argument.predict:
        config.set('eda', 'enabled', False)
        config.set('training', 'enabled', False)
        config.set('predictions', 'enabled', True)

    trainfiles = argument.train_files if argument.train_files else config.get('dataset', 'files')

    models = []

    if trainfiles and not argument.predict:

        ###
        # Step 0. Reading files from command line

        loader = config.get('dataset', 'loader')
        df_train = Plugins.create('loader', loader, config, trainfiles).load()

        ###
        # Step 1. Preprocessing

        # Label is first extracted from the dataset to pass x alone onto the transformers pipeline.
        x_train, y_train, labels = Preprocessors.encodeLabel(config, df_train)
        x_train = Transformers.apply(config, x_train)

        # Rejoin x and y to apply uniform preprocessing to the whole dataset.
        # Global preprocessing comes second, especially for the drop_rows_to_predict_file option.
        df_train = pd.concat(list([x_train, y_train]), axis=1)

        print('\nBefore preprocessing:')
        print(df_train.shape)

        df_train = Preprocessors.apply(config, df_train)

        # Re-separates x and y sets.
        x_train, y_train, encodedLabels = Preprocessors.encodeLabel(config, df_train)

        if config.get('preprocess', 'preprocess_to_file'):

            preprocessed_file = config.get('preprocess', 'preprocess_to_file')
            Files.toCsv(df_train, preprocessed_file)
            print('Preprocessed rows written to: ' + preprocessed_file)

        ###
        # Step 2. EDA plots:

        if config.eq('eda', 'enabled', True):

            identifier = identifier = '/'.join(trainfiles) if len(trainfiles) > 1 else trainfiles
            Plots().run(config, x_train, y_train, labels, identifier)

        ###
        # Step 3. Training:

        if config.eq('training', 'enabled', True):

            models = Training(config, x_train, y_train, labels).run()

    ### Step 4. Predictions from model:

    if config.eq('predictions', 'enabled', True):

        predict_files = argument.predict_files if argument.predict_files else [config.get('predictions', 'predict_file')]
        Predictions(config, predict_files, models).run()

    # @todo
    # Add unit tests
    # Model stacking

if __name__ == "__main__":
    main()