Train-Test Splitting

Endpoint: POST /prepTrainTest

Request Parameters

File Upload

• file (required): CSV file containing the dataset to be split.

Form Parameters

• splitting_method (optional, default: stratified)
  • Method for splitting the dataset.
  • Supported values: stratified, time-based, group-based, shuffle.
• target_column (optional, default: "")
  • The name of the target column used for stratification, grouping, or time-based splitting.
• group_column (optional, default: "")
  • The column to be used for group-based splitting.

Splitting Methods

• Stratified (stratified):

  • Ensures that the training and testing sets maintain the same proportion of target classes.
  • Requires target_column to be specified.

• Time-Based (time-based):

  • Splits data chronologically based on the target_column.
  • Uses the 80th percentile as the threshold for splitting.
  • Requires target_column to contain valid datetime values.

• Group-Based (group-based):

  • Splits data while maintaining group integrity, ensuring all data points in a group remain in the same split.
  • Requires both target_column and group_column to be specified.

• Shuffle (shuffle):

  • Randomly splits the dataset into training and testing sets.
  • Requires target_column to be specified.

Response

• Returns downloadable files containing the split datasets.
• Depending on the splitting method, the returned files will include:
  • X_train.csv, X_test.csv, y_train.csv, y_test.csv (for stratified, group-based, and shuffle methods).
  • train.csv, test.csv (for time-based method).

Errors

• If target_column is not found in the dataset.
• If group_column is specified but does not exist in the dataset (for group-based splitting).
• If time-based splitting is used but target_column is not a valid datetime column.