Model Scores

You can view detailed information about model scores after an experiment is complete by clicking on the Scores option.

Scores link

The Model Scores page that opens includes the following tables:

  • Model and feature tuning leaderboard: This leaderboard shows scoring information based on the scorer that was selected in the experiment. This information is also available in the tuning_leaderboard.json file of the Experiment Summary. You can download that file directly from the bottom of this table.

Model and feature tuning leaderboard

  • Final pipeline scores across cross-validation folds and models: This table shows the final pipeline scores across cross-validation folds and models. Note that if Constant Model was enabled (default), then that model is added in this table as a baseline (reference) only and will be dropped in most cases. This information is also included in the ensemble_base_learner_fold_scores.json file of the Experiment Summary. You can download that file directly from a link at the bottom of this table.

Final pipeline scores

  • Pipeline Description: This shows how the final Stacked Ensemble pipeline was calculated. This information is also included in the ensemble_model_description.json file of the Experiment Summary. You can download that file directly from a link at the bottom of this table.

Final Stacked Ensemble pipeline

  • Final Ensemble Scores: This shows the final scores for each scorer in DAI. If a custom scorer was used in the experiment, that scorer will also appear here. This information is also included in the ensemble_scores.json file of the Experiment Summary. You can download that file directly from a link at the bottom of this table.

Final ensemble scores