def run_number_classifier():
rows = -1 # -1 means retrieving complete set. When testing, set lower for faster training (e.g. 5000).
print('-- Executing number classification')
print('Loading data...')
data, labels = data_retriever.load_mnist(rows)
print('Splitting data...')
training_data, training_labels, test_data, test_labels = split_and_shuffle_data_set(
data, labels)
print('Extracting features...')
extractor = BitmapFeatureExtractor()
extractor.fit(training_data)
training_features = extractor.transform(training_data)
test_features = extractor.transform(test_data)
print('Training classifier...')
classifier = train_classifier(training_features, training_labels)
print('Testing classifier...')
validate_model(classifier,
test_data,
test_features,
test_labels,
bitmap=True)
def run_spam_filter():
row_count = -1 # set this number to some number below 2000 if you are having performance problems
print('-- Executing spam filter')
print('-- Loading data')
data, labels = data_retriever.load_sms(cache_data=False, rows=row_count)
# randomize and split the data
training_data, test_data, training_labels, test_labels = split_and_shuffle_data_set(data, labels)
# fit the transformer
extractor = FeatureExtractor()
extractor.fit(training_data)
# extract the features from the test data
training_features = extractor.transform(training_data)
# train the classifier
classifier = train_classifier(training_features, training_labels)
# generate classification report
test_features = extractor.transform(test_data)
validate_model(classifier, test_data, test_features, test_labels)