Explain Predictions

Explains which features contribute the most to the predictions for the selected instances based on the model and how they contribute.


  • Model: model whose predictions are explained by the widget
  • Background data: dataset needed to compute explanations
  • Data: dataset whose predictions are explained by the widget


  • Selected Data: instances selected from the plot
  • Data: original dataset with an additional column showing whether the instance is selected
  • Scores: SHAP values for each feature. Features that contribute more to prediction have a higher score deviation from 0.

Explain Predictions widget explains classification or regression model’s predictions for the provided data instances.