FAQ¶

Why did I get training outputs but no explanation CSV files?¶

The operator generates global_explanations.csv and local_explanations.csv when generate_explanations: true.

Current behavior:

If generate_explanations is false, no explanation files are written.
Global importance uses a model-derived value when the selected model exposes one and otherwise uses SHAP.
Local explanations use SHAP for every supported model and explain a deterministic sample of up to 200 training rows. base_value plus the signed transformed-feature contributions reconstructs each prediction, subject to rounding.
Keep generate_explanations: true and make sure shap is installed in the operator runtime.

Why did I not get `test_metrics.csv`?¶

test_metrics.csv is written only when:

test_data is provided
the test dataset includes the same feature columns as training
the test dataset also includes target_column

Why does a simple explicit-model run take longer than expected?¶

Explicit models use Optuna-based tuning by default. If you want a faster validation run, set:

model_kwargs:
  tuning_n_trials: 0

Why does `auto` not use my explicit `model_kwargs`?¶

In the current implementation, auto compares the candidate models using its own candidate-selection path and then retrains the selected model. User-supplied explicit-model model_kwargs are not used during the candidate comparison stage.

When should I override `column_types`?¶

Override column_types when automatic inference is likely to be misleading, for example:

identifier-like numeric columns such as customer_id or zip_code
date columns stored as strings
numeric values stored as text with inconsistent formatting

Why did training fail after I disabled preprocessing?¶

The current preprocessing pipeline always converts the final feature matrix to numeric form before fitting. If you disable preprocessing or categorical encoding while raw string categorical columns remain, the matrix may no longer be numeric and model training can fail.