features: Manually setting the validation set for multi-output task

[lizhuoq]

Why are these changes needed?

For original multi-output tasks where the eval_method is holdout, manual setting of the validation set was not possible. This commit introduces a new feature allowing manual setting of the validation set for multi-output tasks.

model = MultiOutputRegressor(
    AutoML(
        task="regression",
        time_budget=1,
        eval_method="holdout",
        multioutput_train_size=len(X_train)
    )
)
model.fit(
    pd.concat([X_train, X_val]),
    pd.concat([y_train, y_val])
)

Related issue number

Checks

• I've included any doc changes needed for https://microsoft.github.io/FLAML/. See https://microsoft.github.io/FLAML/docs/Contribute#documentation to build and test documentation locally.
• I've added tests (if relevant) corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed.

[lizhuoq]

@microsoft-github-policy-service agree

[prdai]

LGTM!

[Copilot]

Pull request overview

This pull request adds support for manually setting a validation set for multi-output tasks when using the "holdout" evaluation method. Previously, users could not manually specify a validation set for multi-output regression tasks. The new multioutput_train_size parameter allows users to concatenate training and validation data and specify where to split them.

Changes:

• Added multioutput_train_size parameter to AutoML class for manual validation set specification
• Implemented _train_val_split method to split concatenated training/validation data
• Added test case demonstrating the new functionality with MultiOutputRegressor

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 4 comments.

File	Description
flaml/automl/automl.py	Added documentation and implementation for the multioutput_train_size parameter, including the split logic in the fit method
test/automl/test_regression.py	Added test_multioutput_train_size function to demonstrate usage of the new feature

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The test function lacks assertions to verify the new multioutput_train_size feature works as expected. Consider adding assertions to validate that the model was trained successfully and that the validation split was performed correctly. For example, you could check that the model produces reasonable predictions or verify internal state that confirms the train/validation split occurred.