Coverage for biobb_ml/neural_networks/classification_neural_network.py: 85%
193 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 ClassificationNeuralNetwork class and the command line interface."""
4import argparse
5import h5py
6import json
7import numpy as np
8import pandas as pd
9from biobb_common.generic.biobb_object import BiobbObject
10from tensorflow.python.keras.saving import hdf5_format
11from sklearn.preprocessing import scale
12from sklearn.model_selection import train_test_split
13from tensorflow.keras import Sequential
14from tensorflow.keras.layers import Dense
15from tensorflow.keras.callbacks import EarlyStopping
16from tensorflow import math
17from biobb_common.configuration import settings
18from biobb_common.tools import file_utils as fu
19from biobb_common.tools.file_utils import launchlogger
20from biobb_ml.neural_networks.common import check_input_path, check_output_path, getHeader, getTargetValue, getFeatures, getIndependentVarsList, getWeight, plotResultsClassMultCM, plotResultsClassBinCM
23class ClassificationNeuralNetwork(BiobbObject):
24 """
25 | biobb_ml ClassificationNeuralNetwork
26 | Wrapper of the TensorFlow Keras Sequential method for classification.
27 | Trains and tests a given dataset and save the complete model for a Neural Network Classification. Visit the `Sequential documentation page <https://www.tensorflow.org/api_docs/python/tf/keras/Sequential>`_ in the TensorFlow Keras official website for further information.
29 Args:
30 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/neural_networks/dataset_classification.csv>`_. Accepted formats: csv (edam:format_3752).
31 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/neural_networks/ref_output_model_classification.h5>`_. Accepted formats: h5 (edam:format_3590).
32 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/neural_networks/ref_output_test_classification.csv>`_. Accepted formats: csv (edam:format_3752).
33 output_plot_path (str) (Optional): Loss, accuracy and MSE plots. File type: output. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/reference/neural_networks/ref_output_plot_classification.png>`_. Accepted formats: png (edam:format_3603).
34 properties (dic - Python dictionary object containing the tool parameters, not input/output files):
35 * **features** (*dict*) - ({}) Independent variables or columns from your dataset you want to train. 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.
36 * **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.
37 * **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 multiple formats, the first one will be picked.
38 * **validation_size** (*float*) - (0.2) [0~1|0.05] Represents the proportion of the dataset to include in the validation split. It should be between 0.0 and 1.0.
39 * **test_size** (*float*) - (0.1) [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.
40 * **hidden_layers** (*list*) - (None) List of dictionaries with hidden layers values. Format: [ { 'size': 50, 'activation': 'relu' } ].
41 * **output_layer_activation** (*string*) - ("softmax") Activation function to use in the output layer. Values: sigmoid (Sigmoid activation function: sigmoid[x] = 1 / [1 + exp[-x]]), tanh (Hyperbolic tangent activation function), relu (Applies the rectified linear unit activation function), softmax (Softmax converts a real vector to a vector of categorical probabilities).
42 * **optimizer** (*string*) - ("Adam") Name of optimizer instance. Values: Adadelta (Adadelta optimization is a stochastic gradient descent method that is based on adaptive learning rate per dimension to address two drawbacks: the continual decay of learning rates throughout training and the need for a manually selected global learning rate), Adagrad (Adagrad is an optimizer with parameter-specific learning rates; which are adapted relative to how frequently a parameter gets updated during training. The more updates a parameter receives; the smaller the updates), Adam (Adam optimization is a stochastic gradient descent method that is based on adaptive estimation of first-order and second-order moments), Adamax (It is a variant of Adam based on the infinity norm. Default parameters follow those provided in the paper. Adamax is sometimes superior to adam; specially in models with embeddings), Ftrl (Optimizer that implements the FTRL algorithm), Nadam (Much like Adam is essentially RMSprop with momentum; Nadam is Adam with Nesterov momentum), RMSprop (Optimizer that implements the RMSprop algorithm), SGD (Gradient descent -with momentum- optimizer).
43 * **learning_rate** (*float*) - (0.02) [0~100|0.01] Determines the step size at each iteration while moving toward a minimum of a loss function
44 * **batch_size** (*int*) - (100) [0~1000|1] Number of samples per gradient update.
45 * **max_epochs** (*int*) - (100) [0~1000|1] Number of epochs to train the model. As the early stopping is enabled, this is a maximum.
46 * **normalize_cm** (*bool*) - (False) Whether or not to normalize the confusion matrix.
47 * **random_state** (*int*) - (5) [1~1000|1] Controls the shuffling applied to the data before applying the split. .
48 * **scale** (*bool*) - (False) Whether or not to scale the input dataset.
49 * **remove_tmp** (*bool*) - (True) [WF property] Remove temporal files.
50 * **restart** (*bool*) - (False) [WF property] Do not execute if output files exist.
52 Examples:
53 This is a use example of how to use the building block from Python::
55 from biobb_ml.neural_networks.classification_neural_network import classification_neural_network
56 prop = {
57 'features': {
58 'columns': [ 'column1', 'column2', 'column3' ]
59 },
60 'target': {
61 'column': 'target'
62 },
63 'validation_size': 0.2,
64 'test_size': .33,
65 'hidden_layers': [
66 {
67 'size': 10,
68 'activation': 'relu'
69 },
70 {
71 'size': 8,
72 'activation': 'relu'
73 }
74 ],
75 'optimizer': 'Adam',
76 'learning_rate': 0.01,
77 'batch_size': 32,
78 'max_epochs': 150
79 }
80 classification_neural_network(input_dataset_path='/path/to/myDataset.csv',
81 output_model_path='/path/to/newModel.h5',
82 output_test_table_path='/path/to/newTable.csv',
83 output_plot_path='/path/to/newPlot.png',
84 properties=prop)
86 Info:
87 * wrapped_software:
88 * name: TensorFlow Keras Sequential
89 * version: >2.1.0
90 * license: MIT
91 * ontology:
92 * name: EDAM
93 * schema: http://edamontology.org/EDAM.owl
95 """
97 def __init__(self, input_dataset_path, output_model_path,
98 output_test_table_path=None, output_plot_path=None, properties=None, **kwargs) -> None:
99 properties = properties or {}
101 # Call parent class constructor
102 super().__init__(properties)
103 self.locals_var_dict = locals().copy()
105 # Input/Output files
106 self.io_dict = {
107 "in": {"input_dataset_path": input_dataset_path},
108 "out": {"output_model_path": output_model_path, "output_test_table_path": output_test_table_path, "output_plot_path": output_plot_path}
109 }
111 # Properties specific for BB
112 self.features = properties.get('features', {})
113 self.target = properties.get('target', {})
114 self.weight = properties.get('weight', {})
115 self.validation_size = properties.get('validation_size', 0.1)
116 self.test_size = properties.get('test_size', 0.1)
117 self.hidden_layers = properties.get('hidden_layers', [])
118 self.output_layer_activation = properties.get('output_layer_activation', 'softmax')
119 self.optimizer = properties.get('optimizer', 'Adam')
120 self.learning_rate = properties.get('learning_rate', 0.02)
121 self.batch_size = properties.get('batch_size', 100)
122 self.max_epochs = properties.get('max_epochs', 100)
123 self.normalize_cm = properties.get('normalize_cm', False)
124 self.random_state = properties.get('random_state', 5)
125 self.scale = properties.get('scale', False)
126 self.properties = properties
128 # Check the properties
129 self.check_properties(properties)
130 self.check_arguments()
132 def check_data_params(self, out_log, err_log):
133 """ Checks all the input/output paths and parameters """
134 self.io_dict["in"]["input_dataset_path"] = check_input_path(self.io_dict["in"]["input_dataset_path"], "input_dataset_path", False, out_log, self.__class__.__name__)
135 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__)
136 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__)
137 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__)
139 def build_model(self, input_shape, output_size):
140 """ Builds Neural network according to hidden_layers property """
142 # create model
143 model = Sequential([])
145 # if no hidden_layers provided, create manually a hidden layer with default values
146 if not self.hidden_layers:
147 self.hidden_layers = [{'size': 50, 'activation': 'relu'}]
149 # generate hidden_layers
150 for i, layer in enumerate(self.hidden_layers):
151 if i == 0:
152 model.add(Dense(layer['size'], activation=layer['activation'], kernel_initializer='he_normal', input_shape=input_shape)) # 1st hidden layer
153 else:
154 model.add(Dense(layer['size'], activation=layer['activation'], kernel_initializer='he_normal'))
156 model.add(Dense(output_size, activation=self.output_layer_activation)) # output layer
158 return model
160 @launchlogger
161 def launch(self) -> int:
162 """Execute the :class:`ClassificationNeuralNetwork <neural_networks.classification_neural_network.ClassificationNeuralNetwork>` neural_networks.classification_neural_network.ClassificationNeuralNetwork object."""
164 # check input/output paths and parameters
165 self.check_data_params(self.out_log, self.err_log)
167 # Setup Biobb
168 if self.check_restart():
169 return 0
170 self.stage_files()
172 # load dataset
173 fu.log('Getting dataset from %s' % self.io_dict["in"]["input_dataset_path"], self.out_log, self.global_log)
174 if 'columns' in self.features:
175 labels = getHeader(self.io_dict["in"]["input_dataset_path"])
176 skiprows = 1
177 else:
178 labels = None
179 skiprows = None
180 data = pd.read_csv(self.io_dict["in"]["input_dataset_path"], header=None, sep="\\s+|;|:|,|\t", engine="python", skiprows=skiprows, names=labels)
182 targets_list = data[getTargetValue(self.target)].to_numpy()
184 X = getFeatures(self.features, data, self.out_log, self.__class__.__name__)
185 fu.log('Features: [%s]' % (getIndependentVarsList(self.features)), self.out_log, self.global_log)
186 # target
187 # y = getTarget(self.target, data, self.out_log, self.__class__.__name__)
188 fu.log('Target: %s' % (str(getTargetValue(self.target))), self.out_log, self.global_log)
189 # weights
190 if self.weight:
191 w = getWeight(self.weight, data, self.out_log, self.__class__.__name__)
193 # shuffle dataset
194 fu.log('Shuffling dataset', self.out_log, self.global_log)
195 shuffled_indices = np.arange(X.shape[0])
196 np.random.shuffle(shuffled_indices)
197 np_X = X.to_numpy()
198 shuffled_X = np_X[shuffled_indices]
199 shuffled_y = targets_list[shuffled_indices]
200 if self.weight:
201 shuffled_w = w[shuffled_indices]
203 # train / test split
204 fu.log('Creating train and test sets', self.out_log, self.global_log)
205 arrays_sets = (shuffled_X, shuffled_y)
206 # if user provide weights
207 if self.weight:
208 arrays_sets = arrays_sets + (shuffled_w,)
209 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)
210 else:
211 X_train, X_test, y_train, y_test = train_test_split(*arrays_sets, test_size=self.test_size, random_state=self.random_state)
213 # scale dataset
214 if self.scale:
215 fu.log('Scaling dataset', self.out_log, self.global_log)
216 X_train = scale(X_train)
218 # build model
219 fu.log('Building model', self.out_log, self.global_log)
220 model = self.build_model((X_train.shape[1],), np.unique(y_train).size)
222 # model summary
223 stringlist = []
224 model.summary(print_fn=lambda x: stringlist.append(x))
225 model_summary = "\n".join(stringlist)
226 fu.log('Model summary:\n\n%s\n' % model_summary, self.out_log, self.global_log)
228 # get optimizer
229 mod = __import__('tensorflow.keras.optimizers', fromlist=[self.optimizer])
230 opt_class = getattr(mod, self.optimizer)
231 opt = opt_class(lr=self.learning_rate)
232 # compile model
233 model.compile(optimizer=opt, loss='sparse_categorical_crossentropy', metrics=['accuracy', 'mse'])
235 # fitting
236 fu.log('Training model', self.out_log, self.global_log)
237 # set an early stopping mechanism
238 # set patience=2, to be a bit tolerant against random validation loss increases
239 early_stopping = EarlyStopping(patience=2)
241 if self.weight:
242 sample_weight = w_train
243 class_weight = []
244 else:
245 # TODO: class_weight not working since TF 2.4.1 update
246 # fu.log('No weight provided, class_weight will be estimated from the target data', self.out_log, self.global_log)
247 fu.log('No weight provided', self.out_log, self.global_log)
248 sample_weight = None
249 class_weight = [] # compute_class_weight('balanced', np.unique(y_train), y_train)
251 print(class_weight)
252 # fit the model
253 mf = model.fit(X_train,
254 y_train,
255 class_weight=class_weight,
256 sample_weight=sample_weight,
257 batch_size=self.batch_size,
258 epochs=self.max_epochs,
259 callbacks=[early_stopping],
260 validation_split=self.validation_size,
261 verbose=1)
263 fu.log('Total epochs performed: %s' % len(mf.history['loss']), self.out_log, self.global_log)
265 train_metrics = pd.DataFrame()
266 train_metrics['metric'] = ['Train loss', ' Train accuracy', 'Train MSE', 'Validation loss', 'Validation accuracy', 'Validation MSE']
267 train_metrics['coefficient'] = [mf.history['loss'][-1], mf.history['accuracy'][-1], mf.history['mse'][-1], mf.history['val_loss'][-1], mf.history['val_accuracy'][-1], mf.history['val_mse'][-1]]
269 fu.log('Training metrics\n\nTRAINING METRICS TABLE\n\n%s\n' % train_metrics, self.out_log, self.global_log)
271 # confusion matrix
272 train_predictions = model.predict(X_train)
273 train_predictions = np.around(train_predictions, decimals=2)
274 norm_pred = []
275 [norm_pred.append(np.argmax(pred, axis=0)) for pred in train_predictions]
276 cnf_matrix_train = math.confusion_matrix(y_train, norm_pred).numpy()
277 np.set_printoptions(precision=2)
278 if self.normalize_cm:
279 cnf_matrix_train = cnf_matrix_train.astype('float') / cnf_matrix_train.sum(axis=1)[:, np.newaxis]
280 cm_type = 'NORMALIZED CONFUSION MATRIX'
281 else:
282 cm_type = 'CONFUSION MATRIX, WITHOUT NORMALIZATION'
284 fu.log('Calculating confusion matrix for training dataset\n\n%s\n\n%s\n' % (cm_type, cnf_matrix_train), self.out_log, self.global_log)
286 # testing
287 if self.scale:
288 X_test = scale(X_test)
289 fu.log('Testing model', self.out_log, self.global_log)
290 test_loss, test_accuracy, test_mse = model.evaluate(X_test, y_test)
292 test_metrics = pd.DataFrame()
293 test_metrics['metric'] = ['Test loss', ' Test accuracy', 'Test MSE']
294 test_metrics['coefficient'] = [test_loss, test_accuracy, test_mse]
296 fu.log('Testing metrics\n\nTESTING METRICS TABLE\n\n%s\n' % test_metrics, self.out_log, self.global_log)
298 # predict data from X_test
299 test_predictions = model.predict(X_test)
300 test_predictions = np.around(test_predictions, decimals=2)
301 tpr = tuple(map(tuple, test_predictions))
303 test_table = pd.DataFrame()
304 test_table['P' + np.array2string(np.unique(y_train))] = tpr
305 test_table['target'] = y_test
307 fu.log('TEST DATA\n\n%s\n' % test_table, self.out_log, self.global_log)
309 # confusion matrix
310 norm_pred = []
311 [norm_pred.append(np.argmax(pred, axis=0)) for pred in test_predictions]
312 cnf_matrix_test = math.confusion_matrix(y_test, norm_pred).numpy()
313 np.set_printoptions(precision=2)
314 if self.normalize_cm:
315 cnf_matrix_test = cnf_matrix_test.astype('float') / cnf_matrix_test.sum(axis=1)[:, np.newaxis]
316 cm_type = 'NORMALIZED CONFUSION MATRIX'
317 else:
318 cm_type = 'CONFUSION MATRIX, WITHOUT NORMALIZATION'
320 fu.log('Calculating confusion matrix for testing dataset\n\n%s\n\n%s\n' % (cm_type, cnf_matrix_test), self.out_log, self.global_log)
322 # save test data
323 if (self.io_dict["out"]["output_test_table_path"]):
324 fu.log('Saving testing data to %s' % self.io_dict["out"]["output_test_table_path"], self.out_log, self.global_log)
325 test_table.to_csv(self.io_dict["out"]["output_test_table_path"], index=False, header=True)
327 # create test plot
328 if (self.io_dict["out"]["output_plot_path"]):
329 vs = np.unique(targets_list)
330 vs.sort()
331 if len(vs) > 2:
332 plot = plotResultsClassMultCM(mf.history, cnf_matrix_train, cnf_matrix_test, self.normalize_cm, vs)
333 fu.log('Saving confusion matrix plot to %s' % self.io_dict["out"]["output_plot_path"], self.out_log, self.global_log)
334 else:
335 plot = plotResultsClassBinCM(mf.history, train_predictions, test_predictions, y_train, y_test, cnf_matrix_train, cnf_matrix_test, self.normalize_cm, vs)
336 fu.log('Saving binary classifier evaluator plot to %s' % self.io_dict["out"]["output_plot_path"], self.out_log, self.global_log)
337 plot.savefig(self.io_dict["out"]["output_plot_path"], dpi=150)
339 # save model and parameters
340 vs = np.unique(targets_list)
341 vs.sort()
342 vars_obj = {
343 'features': self.features,
344 'target': self.target,
345 'scale': self.scale,
346 'vs': vs.tolist(),
347 'type': 'classification'
348 }
349 variables = json.dumps(vars_obj)
350 fu.log('Saving model to %s' % self.io_dict["out"]["output_model_path"], self.out_log, self.global_log)
351 with h5py.File(self.io_dict["out"]["output_model_path"], mode='w') as f:
352 hdf5_format.save_model_to_hdf5(model, f)
353 f.attrs['variables'] = variables
355 # Copy files to host
356 self.copy_to_host()
358 self.tmp_files.extend([
359 self.stage_io_dict.get("unique_dir")
360 ])
361 self.remove_tmp_files()
363 self.check_arguments(output_files_created=True, raise_exception=False)
365 return 0
368def classification_neural_network(input_dataset_path: str, output_model_path: str, output_test_table_path: str = None, output_plot_path: str = None, properties: dict = None, **kwargs) -> int:
369 """Execute the :class:`AutoencoderNeuralNetwork <neural_networks.classification_neural_network.AutoencoderNeuralNetwork>` class and
370 execute the :meth:`launch() <neural_networks.classification_neural_network.AutoencoderNeuralNetwork.launch>` method."""
372 return ClassificationNeuralNetwork(input_dataset_path=input_dataset_path,
373 output_model_path=output_model_path,
374 output_test_table_path=output_test_table_path,
375 output_plot_path=output_plot_path,
376 properties=properties, **kwargs).launch()
379def main():
380 """Command line execution of this building block. Please check the command line documentation."""
381 parser = argparse.ArgumentParser(description="Wrapper of the TensorFlow Keras Sequential method.", formatter_class=lambda prog: argparse.RawTextHelpFormatter(prog, width=99999))
382 parser.add_argument('--config', required=False, help='Configuration file')
384 # Specific args of each building block
385 required_args = parser.add_argument_group('required arguments')
386 required_args.add_argument('--input_dataset_path', required=True, help='Path to the input dataset. Accepted formats: csv.')
387 required_args.add_argument('--output_model_path', required=True, help='Path to the output model file. Accepted formats: h5.')
388 parser.add_argument('--output_test_table_path', required=False, help='Path to the test table file. Accepted formats: csv.')
389 parser.add_argument('--output_plot_path', required=False, help='Loss, accuracy and MSE plots. Accepted formats: png.')
391 args = parser.parse_args()
392 args.config = args.config or "{}"
393 properties = settings.ConfReader(config=args.config).get_prop_dic()
395 # Specific call of each building block
396 classification_neural_network(input_dataset_path=args.input_dataset_path,
397 output_model_path=args.output_model_path,
398 output_test_table_path=args.output_test_table_path,
399 output_plot_path=args.output_plot_path,
400 properties=properties)
403if __name__ == '__main__':
404 main()