def __generate_data(indexs, labels, features, positive_rate):
"""
generate data set according to the `indexs` and `positive_rate`
:param indexs: indexs which will select data from raw data set
:param labels: all labels of raw data set
:param features: feature matrix
:param positive_rate: positive_rate in data set
:return: feature matrix, labels, balanced indexs
"""
# balance the data set
balanced_indexs = Feature.balance_index(indexs, labels, positive_rate)
# sample labels
labels = [labels[index] for index in balanced_indexs]
# sample features
features = Feature.sample_row(features, balanced_indexs)
return features, labels, balanced_indexs
def __generate_data(indexs, labels, features, positive_rate):
"""
generate data set according to the `indexs` and `positive_rate`
:param indexs: indexs which will select data from raw data set
:param labels: all labels of raw data set
:param features: feature matrix
:param positive_rate: positive_rate in data set
:return: feature matrix, labels, balanced indexs
"""
# balance the data set
balanced_indexs = Feature.balance_index(indexs, labels, positive_rate)
# sample labels
labels = [labels[index] for index in balanced_indexs]
# sample features
features = Feature.sample_row(features, balanced_indexs)
return features, labels, balanced_indexs
