def train(self, path, database):
"""
Trains an specific model with the features and labels read from a file.
Parameters
----------
path: string
The location of the file with the features and labels for training.
database: string
The name of the database.
Return
----------
succesful: boolean
Wether the training of the model was succesful.
"""
if not Utilities.checkFile(path, database):
return False
else:
dataframe = Utilities.readCSV(path)
labels = dataframe['Diagnosis'].values
classes, ids = Utilities.getClasses(labels)
encoded_labels = np_utils.to_categorical(ids, len(classes))
if database == 'Caldas':
features = dataframe[list(['CDR', 'MMSE', 'MoCA', 'GDS'])].values
self.classifier_caldas.train(features, encoded_labels)
self.classifier_caldas.trained = True
self.classifier_caldas.classes = classes
else:
features = dataframe[list(['CDR', 'MMSE', 'Age', 'Education'])].values
self.classifier_adni.train(features, encoded_labels)
self.classifier_adni.trained = True
self.classifier_adni.classes = classes
return True
def train(self, features, labels):
"""
Train the model itself with a data set (features, labels).
The training method uses bootstrapping to extract samples from the data set.
Parameters
----------
features: array-like of shape = [number_samples, number_features]
The training input samples.
labels: array-like of shape = [number_samples]
The target values (class labels in classification).
Return
----------
None
"""
number_samples, number_features = features.shape
max_samples = number_samples
random_state = check_random_state(None)
seeds = random_state.randint(numpy.iinfo(numpy.int32).max,
size=self.number_classifiers)
classes, ids = Utilities.getClasses(labels)
encoded_labels = np_utils.to_categorical(ids, len(classes))
for i in range(self.number_classifiers):
random_state = check_random_state(seeds[i])
indices = random_state.randint(0, number_samples, max_samples)
if i % 2 != 0:
self.model[i].train(features[indices], labels[indices])
self.model[i].classes = self.model[i].model.classes_
else:
self.model[i].train(features[indices], encoded_labels[indices])
self.model[i].classes = classes
self.classes = classes
def validate(self, path, number_folds, database):
"""
Compute a model's performance metrics based on k-fold cross-validation technique.
Parameters
----------
path: string
The location of the file with the features and labels for validation.
number_folds: int
The amount of folds for the k-fold cross-validation.
database: string
The name of the database.
Return
----------
checked: boolean
Whether the columns/classes in the file correspond to the ones associated with the database.
validated: boolean
Wether the validation of the model was succesful.
accuracy: float
The accuracy of the model based on it's confusion matrix.
precision: float
The precision of the model based on it's confusion matrix.
sensitivity: float
The sensitivity of the model based on it's confusion matrix.
specificity: float
The specificity of the model based on it's confusion matrix.
kappa: float
The Cohen's Kappa of the model based on it's confusion matrix.
"""
if not Utilities.checkFile(path, database):
return False, True, None, None, None, None, None
else:
dataframe = Utilities.readCSV(path)
labels = dataframe['Diagnosis'].values
classes, ids = Utilities.getClasses(labels)
encoded_labels = np_utils.to_categorical(ids, len(classes))
if database == 'Caldas':
if self.classifier_caldas.trained:
features = dataframe[list(['CDR', 'MMSE', 'MoCA', 'GDS'])].values
accuracy, precision, sensitivity, specificity, kappa = self.classifier_caldas.validate(features, labels, number_folds, encoded_labels)
else:
return True, False, None, None, None, None, None
else:
if self.classifier_adni.trained:
features = dataframe[list(['CDR', 'MMSE', 'Age', 'Education'])].values
accuracy, precision, sensitivity, specificity, kappa = self.classifier_adni.validate(features, labels, number_folds, encoded_labels)
else:
return True, False, None, None, None, None, None
return True, True, accuracy, precision, sensitivity, specificity, kappa