Enhance train_val_test_split with Flexible Test Dataset Support

[infingardi]

What was changed

Changed function train_val_test_split in _bibmon_tools.py

• Added two more optionals parameters data_test and start_test.

Objective

Enable the user to train and validate a model with one dataset and test it with another, allowing flexible partitioning for training, validation, and testing phases.

Importance

The new parameters data_test and start_test improve the function train_val_test_split by allowing the user to train and validate a model with one dataset and test it with another. This flexibility enables testing on data from different distributions or scenarios, enhancing model evaluation. The start_test parameter further allows for precise control over the test set's starting point within the provided test data, making the function more versatile for time-series and other partitioned datasets.

How to use

import bibmon

(X_train, X_validation, 
 X_test, Y_train, 
 Y_validation, Y_test) = bibmon.train_val_test_split(dataNormalOperation, 
                                                     start_train = start_train,
                                                     end_train = end_train, 
                                                     end_validation = end_validation,
                                                     end_test = end_test,
                                                     tags_Y = ['T-TPT'],
                                                     data_test = dataFault,
                                                     start_test = start_test)