TabularPredictor.learning_curves

TabularPredictor.learning_curves() → tuple[dict, dict]

Retrieves learning curves generated during predictor.fit(). Will not work if the learning_curves flag was not set during training. Note that learning curves are only generated for iterative learners with learning curve support.

Parameters:

None

Returns:

• metadata (dict) – A dictionary containing metadata related to the training process.

• model_data (dict) – A dictionary containing the learning curves across all models. To see curve_data format, refer to AbstractModel’s save_learning_curves() method.

  {

  “model”: curve_data, “model”: curve_data, “model”: curve_data, “model”: curve_data,

  }