def scale(dataset, with_mean=True):
scaled_features = skscale(dataset.all.features, with_mean=with_mean)
training_size = dataset.training_size
training = Data(scaled_features[:training_size, :],
dataset.all.target[:training_size])
testing = Data(scaled_features[training_size:, :],
dataset.all.target[training_size:])
return DataSet(training, testing)
def feature_transformation(features, preprocessing='normalization'):
n_samples, n_features = features.shape
if preprocessing == 'scale':
features = skscale(features, copy=False)
elif preprocessing == 'minmax':
minmax_scale = MinMaxScaler().fit(features)
features = minmax_scale.transform(features)
elif preprocessing == 'normalization':
features = np.sqrt(n_features) * normalize(features, copy=False)
else:
print('No preprocessing is applied')
return features