Coverage for biobb_ml/regression/linear_regression.py: 87%
155 statements
« prev ^ index » next coverage.py v7.5.1, created at 2024-05-07 09:39 +0000
1#!/usr/bin/env python3
3"""Module containing the LinearRegression class and the command line interface."""
4import argparse
5import numpy as np
6import pandas as pd
7import joblib
8from biobb_common.generic.biobb_object import BiobbObject
9from sklearn.preprocessing import StandardScaler
10from sklearn.model_selection import train_test_split
11from sklearn.feature_selection import f_regression
12from sklearn.metrics import mean_squared_error, r2_score
13from sklearn import linear_model
14from biobb_common.configuration import settings
15from biobb_common.tools import file_utils as fu
16from biobb_common.tools.file_utils import launchlogger
17from biobb_ml.regression.common import check_input_path, check_output_path, getHeader, getIndependentVars, getIndependentVarsList, getTarget, getTargetValue, getWeight, adjusted_r2, plotResults
20class LinearRegression(BiobbObject):
21 """
22 | biobb_ml LinearRegression
23 | Wrapper of the scikit-learn LinearRegression method.
24 | Trains and tests a given dataset and saves the model and scaler. Visit the `LinearRegression documentation page <https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html>`_ in the sklearn official website for further information.
26 Args:
27 input_dataset_path (str): Path to the input dataset. File type: input. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/data/regression/dataset_linear_regression.csv>`_. Accepted formats: csv (edam:format_3752).
28 output_model_path (str): Path to the output model file. File type: output. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/reference/regression/ref_output_model_linear_regression.pkl>`_. Accepted formats: pkl (edam:format_3653).
29 output_test_table_path (str) (Optional): Path to the test table file. File type: output. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/reference/regression/ref_output_test_linear_regression.csv>`_. Accepted formats: csv (edam:format_3752).
30 output_plot_path (str) (Optional): Residual plot checks the error between actual values and predicted values. File type: output. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/reference/regression/ref_output_plot_linear_regression.png>`_. Accepted formats: png (edam:format_3603).
31 properties (dic - Python dictionary object containing the tool parameters, not input/output files):
32 * **independent_vars** (*dict*) - ({}) Independent variables you want to train from your dataset. You can specify either a list of columns names from your input dataset, a list of columns indexes or a range of columns indexes. Formats: { "columns": ["column1", "column2"] } or { "indexes": [0, 2, 3, 10, 11, 17] } or { "range": [[0, 20], [50, 102]] }. In case of mulitple formats, the first one will be picked.
33 * **target** (*dict*) - ({}) Dependent variable you want to predict from your dataset. You can specify either a column name or a column index. Formats: { "column": "column3" } or { "index": 21 }. In case of mulitple formats, the first one will be picked.
34 * **weight** (*dict*) - ({}) Weight variable from your dataset. You can specify either a column name or a column index. Formats: { "column": "column3" } or { "index": 21 }. In case of mulitple formats, the first one will be picked.
35 * **random_state_train_test** (*int*) - (5) [1~1000|1] Controls the shuffling applied to the data before applying the split.
36 * **test_size** (*float*) - (0.2) [0~1|0.05] Represents the proportion of the dataset to include in the test split. It should be between 0.0 and 1.0.
37 * **scale** (*bool*) - (False) Whether or not to scale the input dataset.
38 * **remove_tmp** (*bool*) - (True) [WF property] Remove temporal files.
39 * **restart** (*bool*) - (False) [WF property] Do not execute if output files exist.
41 Examples:
42 This is a use example of how to use the building block from Python::
44 from biobb_ml.regression.linear_regression import linear_regression
45 prop = {
46 'independent_vars': {
47 'columns': [ 'column1', 'column2', 'column3' ]
48 },
49 'target': {
50 'column': 'target'
51 },
52 'test_size': 0.2
53 }
54 linear_regression(input_dataset_path='/path/to/myDataset.csv',
55 output_model_path='/path/to/newModel.pkl',
56 output_test_table_path='/path/to/newTable.csv',
57 output_plot_path='/path/to/newPlot.png',
58 properties=prop)
60 Info:
61 * wrapped_software:
62 * name: scikit-learn LinearRegression
63 * version: >=0.24.2
64 * license: BSD 3-Clause
65 * ontology:
66 * name: EDAM
67 * schema: http://edamontology.org/EDAM.owl
69 """
71 def __init__(self, input_dataset_path, output_model_path,
72 output_test_table_path=None, output_plot_path=None, properties=None, **kwargs) -> None:
73 properties = properties or {}
75 # Call parent class constructor
76 super().__init__(properties)
77 self.locals_var_dict = locals().copy()
79 # Input/Output files
80 self.io_dict = {
81 "in": {"input_dataset_path": input_dataset_path},
82 "out": {"output_model_path": output_model_path, "output_test_table_path": output_test_table_path, "output_plot_path": output_plot_path}
83 }
85 # Properties specific for BB
86 self.independent_vars = properties.get('independent_vars', {})
87 self.target = properties.get('target', {})
88 self.weight = properties.get('weight', {})
89 self.random_state_train_test = properties.get('random_state_train_test', 5)
90 self.test_size = properties.get('test_size', 0.2)
91 self.scale = properties.get('scale', False)
92 self.properties = properties
94 # Check the properties
95 self.check_properties(properties)
96 self.check_arguments()
98 def check_data_params(self, out_log, err_log):
99 """ Checks all the input/output paths and parameters """
100 self.io_dict["in"]["input_dataset_path"] = check_input_path(self.io_dict["in"]["input_dataset_path"], "input_dataset_path", out_log, self.__class__.__name__)
101 self.io_dict["out"]["output_model_path"] = check_output_path(self.io_dict["out"]["output_model_path"], "output_model_path", False, out_log, self.__class__.__name__)
102 if self.io_dict["out"]["output_test_table_path"]:
103 self.io_dict["out"]["output_test_table_path"] = check_output_path(self.io_dict["out"]["output_test_table_path"], "output_test_table_path", True, out_log, self.__class__.__name__)
104 if self.io_dict["out"]["output_plot_path"]:
105 self.io_dict["out"]["output_plot_path"] = check_output_path(self.io_dict["out"]["output_plot_path"], "output_plot_path", True, out_log, self.__class__.__name__)
107 @launchlogger
108 def launch(self) -> int:
109 """Execute the :class:`LinearRegression <regression.linear_regression.LinearRegression>` regression.linear_regression.LinearRegression object."""
111 # check input/output paths and parameters
112 self.check_data_params(self.out_log, self.err_log)
114 # Setup Biobb
115 if self.check_restart():
116 return 0
117 self.stage_files()
119 # load dataset
120 fu.log('Getting dataset from %s' % self.io_dict["in"]["input_dataset_path"], self.out_log, self.global_log)
121 if 'columns' in self.independent_vars:
122 labels = getHeader(self.io_dict["in"]["input_dataset_path"])
123 skiprows = 1
124 else:
125 labels = None
126 skiprows = None
127 data = pd.read_csv(self.io_dict["in"]["input_dataset_path"], header=None, sep="\\s+|;|:|,|\t", engine="python", skiprows=skiprows, names=labels)
129 # declare inputs, targets and weights
130 # the inputs are all the independent variables
131 X = getIndependentVars(self.independent_vars, data, self.out_log, self.__class__.__name__)
132 fu.log('Independent variables: [%s]' % (getIndependentVarsList(self.independent_vars)), self.out_log, self.global_log)
133 # target
134 y = getTarget(self.target, data, self.out_log, self.__class__.__name__)
135 fu.log('Target: %s' % (getTargetValue(self.target)), self.out_log, self.global_log)
136 # weights
137 if self.weight:
138 w = getWeight(self.weight, data, self.out_log, self.__class__.__name__)
139 fu.log('Weight column provided', self.out_log, self.global_log)
141 # train / test split
142 fu.log('Creating train and test sets', self.out_log, self.global_log)
143 arrays_sets = (X, y)
144 # if user provide weights
145 if self.weight:
146 arrays_sets = arrays_sets + (w,)
147 X_train, X_test, y_train, y_test, w_train, w_test = train_test_split(*arrays_sets, test_size=self.test_size, random_state=self.random_state_train_test)
148 else:
149 X_train, X_test, y_train, y_test = train_test_split(*arrays_sets, test_size=self.test_size, random_state=self.random_state_train_test)
151 # scale dataset
152 if self.scale:
153 fu.log('Scaling dataset', self.out_log, self.global_log)
154 scaler = StandardScaler()
155 X_train = scaler.fit_transform(X_train)
157 # regression
158 fu.log('Training dataset applying linear regression', self.out_log, self.global_log)
159 model = linear_model.LinearRegression()
160 arrays_fit = (X_train, y_train)
161 # if user provide weights
162 if self.weight:
163 arrays_fit = arrays_fit + (w_train,)
165 model.fit(*arrays_fit)
167 # scores and coefficients train
168 y_hat_train = model.predict(X_train)
169 rmse = (np.sqrt(mean_squared_error(y_train, y_hat_train)))
170 rss = ((y_train - y_hat_train) ** 2).sum()
171 score = r2_score(y_hat_train, y_train)
172 bias = model.intercept_
173 coef = model.coef_
174 coef = ['%.3f' % item for item in coef]
175 adj_r2 = adjusted_r2(X_train, y_train, score)
176 p_values = f_regression(X_train, y_train)[1]
177 p_values = ['%.3f' % item for item in p_values]
179 # r-squared
180 r2_table = pd.DataFrame()
181 r2_table["feature"] = ['R2', 'Adj. R2', 'RMSE', 'RSS']
182 r2_table['coefficient'] = [score, adj_r2, rmse, rss]
184 # p-values
185 cols = ['bias']
186 cols.extend(list(getIndependentVarsList(self.independent_vars).split(", ")))
187 coefs_table = pd.DataFrame(cols, columns=['feature'])
188 c = [round(bias, 3)]
189 c.extend(coef)
190 c = list(map(float, c))
191 coefs_table['coefficient'] = c
192 p = [0]
193 p.extend(p_values)
195 coefs_table['p-value'] = p
196 fu.log('Calculating scores and coefficients for training dataset\n\nR2, ADJUSTED R2 & RMSE\n\n%s\n\nCOEFFS & P-VALUES\n\n%s\n' % (r2_table, coefs_table), self.out_log, self.global_log)
198 # testing
199 # predict data from x_test
200 if self.scale:
201 X_test = scaler.transform(X_test)
202 y_hat_test = model.predict(X_test)
203 test_table = pd.DataFrame(y_hat_test, columns=['prediction'])
204 # reset y_test (problem with old indexes column)
205 y_test = y_test.reset_index(drop=True)
206 # add real values to predicted ones in test_table table
207 test_table['target'] = y_test
208 # calculate difference between target and prediction (absolute and %)
209 test_table['residual'] = test_table['target'] - test_table['prediction']
210 test_table['difference %'] = np.absolute(test_table['residual']/test_table['target']*100)
211 pd.set_option('display.float_format', lambda x: '%.2f' % x)
212 # sort by difference in %
213 test_table = test_table.sort_values(by=['difference %'])
214 test_table = test_table.reset_index(drop=True)
215 fu.log('Testing\n\nTEST DATA\n\n%s\n' % test_table, self.out_log, self.global_log)
217 # scores and coefficients test
218 r2_test = r2_score(y_hat_test, y_test)
219 adj_r2_test = adjusted_r2(X_test, y_test, r2_test)
220 rmse_test = np.sqrt(mean_squared_error(y_test, y_hat_test))
221 rss_test = ((y_test - y_hat_test) ** 2).sum()
223 # r-squared
224 pd.set_option('display.float_format', lambda x: '%.6f' % x)
225 r2_table_test = pd.DataFrame()
226 r2_table_test["feature"] = ['R2', 'Adj. R2', 'RMSE', 'RSS']
227 r2_table_test['coefficient'] = [r2_test, adj_r2_test, rmse_test, rss_test]
229 fu.log('Calculating scores and coefficients for testing dataset\n\nR2, ADJUSTED R2 & RMSE\n\n%s\n' % r2_table_test, self.out_log, self.global_log)
231 if (self.io_dict["out"]["output_test_table_path"]):
232 fu.log('Saving testing data to %s' % self.io_dict["out"]["output_test_table_path"], self.out_log, self.global_log)
233 test_table.to_csv(self.io_dict["out"]["output_test_table_path"], index=False, header=True)
235 # create test plot
236 if (self.io_dict["out"]["output_plot_path"]):
237 fu.log('Saving residual plot to %s' % self.io_dict["out"]["output_plot_path"], self.out_log, self.global_log)
238 y_hat_test = y_hat_test.flatten()
239 y_hat_train = y_hat_train.flatten()
240 plot = plotResults(y_train, y_hat_train, y_test, y_hat_test)
241 plot.savefig(self.io_dict["out"]["output_plot_path"], dpi=150)
243 # save model, scaler and parameters
244 variables = {
245 'target': self.target,
246 'independent_vars': self.independent_vars,
247 'scale': self.scale
248 }
249 fu.log('Saving model to %s' % self.io_dict["out"]["output_model_path"], self.out_log, self.global_log)
250 with open(self.io_dict["out"]["output_model_path"], "wb") as f:
251 joblib.dump(model, f)
252 if self.scale:
253 joblib.dump(scaler, f)
254 joblib.dump(variables, f)
256 # Copy files to host
257 self.copy_to_host()
259 self.tmp_files.extend([
260 self.stage_io_dict.get("unique_dir")
261 ])
262 self.remove_tmp_files()
264 self.check_arguments(output_files_created=True, raise_exception=False)
266 return 0
269def linear_regression(input_dataset_path: str, output_model_path: str, output_test_table_path: str = None, output_plot_path: str = None, properties: dict = None, **kwargs) -> int:
270 """Execute the :class:`LinearRegression <regression.linear_regression.LinearRegression>` class and
271 execute the :meth:`launch() <regression.linear_regression.LinearRegression.launch>` method."""
273 return LinearRegression(input_dataset_path=input_dataset_path,
274 output_model_path=output_model_path,
275 output_test_table_path=output_test_table_path,
276 output_plot_path=output_plot_path,
277 properties=properties, **kwargs).launch()
280def main():
281 """Command line execution of this building block. Please check the command line documentation."""
282 parser = argparse.ArgumentParser(description="Wrapper of the scikit-learn LinearRegression method.", formatter_class=lambda prog: argparse.RawTextHelpFormatter(prog, width=99999))
283 parser.add_argument('--config', required=False, help='Configuration file')
285 # Specific args of each building block
286 required_args = parser.add_argument_group('required arguments')
287 required_args.add_argument('--input_dataset_path', required=True, help='Path to the input dataset. Accepted formats: csv.')
288 required_args.add_argument('--output_model_path', required=True, help='Path to the output model file. Accepted formats: pkl.')
289 parser.add_argument('--output_test_table_path', required=False, help='Path to the test table file. Accepted formats: csv.')
290 parser.add_argument('--output_plot_path', required=False, help='Residual plot checks the error between actual values and predicted values. Accepted formats: png.')
292 args = parser.parse_args()
293 args.config = args.config or "{}"
294 properties = settings.ConfReader(config=args.config).get_prop_dic()
296 # Specific call of each building block
297 linear_regression(input_dataset_path=args.input_dataset_path,
298 output_model_path=args.output_model_path,
299 output_test_table_path=args.output_test_table_path,
300 output_plot_path=args.output_plot_path,
301 properties=properties)
304if __name__ == '__main__':
305 main()