Feature: Optionally Include Training Data in SklearnSerializer Output

[Gnpd]

Feature: Optionally Include Training Data in SklearnSerializer Output

Summary:
Add support for optionally including training data (X, y) in the output of SklearnSerializer.serialize. This would allow users to pass the original training features and targets when serializing a model, resulting in a training_data key in the serialized dictionary.

Motivation:

• Enables reproducibility and easier model inspection by storing the data used for training alongside the model parameters and attributes.
• Facilitates downstream tasks such as model validation, auditing, and sharing, where access to the original training data is beneficial.

Proposed API Change:

• Update the serialize method to accept optional X and y parameters.
• If provided, include a training_data key in the output dictionary:

  {
      ...existing keys...,
      "training_data": {
          "X": <serialized X>,
          "y": <serialized y>
      }
  }

• If not provided, omit the training_data key.

Example Usage:

serializer = SklearnSerializer()
serialized = serializer.serialize(model, X=X_train, y=y_train)

Notes:

• Training data should be serialized using the existing conversion utilities to ensure compatibility with JSON and other formats.
• This addition should be fully optional and backward compatible.