Split a dataset-Ant Privacy-Preserving Computation Platform-阿里云帮助中心

This topic describes how to use the API to split a dataset into a training dataset and a test dataset.

Function path

fascia.data.horizontal.dataframe.train_test_split

Function definition

def train_test_split(data: HDataFrame, 
                     ratio: float, 
                     random_state: int = None, 
                     shuffle: bool = True) -> (HDataFrame, HDataFrame):

Parameters

Parameter	Type	Description
data	HDataFrame	The federated dataset to split.
ratio	Float	The split ratio. The value must be between 0 and 1, inclusive. The value can be accurate to three decimal places.
random_state	Integer	The random number seed. If specified, the split result is consistent for the same seed. The default value is None.
shuffle	Bool	Specifies whether to shuffle the data. The default value is True.

Example

from fascia.data.horizontal.dataframe import train_test_split
# Split an existing federated dataset and save the two resulting datasets.
# Assume that fed_df is a pre-existing federated dataset.
train_set, test_set = train_test_split(fed_df, 0.7) 
save_fed_dataframe(train_set, '$output1')
save_fed_dataframe(test_set, '$output2')

Return value

A tuple containing two federated tables.