TabularPredictor.model_failures

TabularPredictor.model_failures(verbose: bool = False) → DataFrame

[Advanced] Get the model failures that occurred during the fitting of this model, in the form of a pandas DataFrame.

This is useful for in-depth debugging of model failures and identifying bugs.

For more information on model failures, refer to predictor.info()[‘model_info_failures’]

Parameters:

verbose (bool, default = False) – If True, the output DataFrame is printed to stdout.

Returns:

model_failures_df – A DataFrame of model failures. Each row corresponds to a model failure, and columns correspond to meta information about that model.

Included Columns:

”model”: The name of the model that failed “exc_type”: The class name of the exception raised “total_time”: The total time in seconds taken by the model prior to the exception (lost time due to the failure) “model_type”: The class name of the model “child_model_type”: The child class name of the model “is_initialized” “is_fit” “is_valid” “can_infer” “num_features” “num_models” “memory_size” “hyperparameters” “hyperparameters_fit” “child_hyperparameters” “child_hyperparameters_fit” “exc_str”: The string message contained in the raised exception “exc_traceback”: The full traceback message of the exception as a string “exc_order”: The order of the model failure (starting from 1)

Return type:

pd.DataFrame