Completed Experiment

Completed Experiment Actions

After an experiment status changes from RUNNING to COMPLETE, the UI provides you with several actions that you can perform:

Experiment Complete menu

  • Deploy (Local and Cloud): Refer to Deploying the MOJO Pipeline.

  • Interpret this Model: Refer to Interpreting a Model. (Not supported for Image or multiclass Time Series experiments.)

  • Diagnose Model on New Dataset: Refer to diagnosing_a_model.

  • Model Actions dropdown:

    • Predict: Refer to Scoring on Another Dataset.

    • Transform Dataset: Refer to Transforming Another Dataset. (Not available for Time Series experiments.)

    • Shapley Values dropdown: Download Shapley values for original or transformed features. Driverless AI calls XGBoost and LightGBM SHAP functions to get contributions for transformed features. Shapley for original features is approximated from transformed features by evenly splitting (naive method) the contribution amongst the contributing features. See Shapley values for details. Select Fast Approximation to make Shapley predictions using only a single fold and model from all of the available folds and models in the ensemble. For more information on the fast approximation options, refer to the fast_approx_num_trees and fast_approx_do_one_fold_one_model config.toml settings.

      • Original Features (Fast Approximation)

      • Original Features

      • Transformed Features (Fast Approximation)

      • Transformed Features

  • Download Predictions dropdown:

    • Training (Holdout) Predictions: In csv format, available if a validation set was NOT provided.

    • Validation Set Predictions: In csv format, available if a validation set was provided.

    • Test Set Predictions: In csv format, available if a test dataset is used.

  • Download Python Scoring Pipeline: A standalone Python scoring pipeline for H2O Driverless AI. Refer to Driverless AI Standalone Python Scoring Pipeline.

  • Download MOJO Scoring Pipeline: A standalone Model Object, Optimized scoring pipeline. Refer to MOJO Scoring Pipelines. (Not available for TensorFlow or RuleFit models.)

  • Visualize Scoring Pipeline (Experimental): Opens an experiment pipeline visualization page. Refer to Visualizing the Scoring Pipeline.

  • Download Summary & Logs: A zip file containing the following files. Refer to the Experiment Summary section for more information.

    • Experiment logs (regular and anonymized)

    • A summary of the experiment

    • The experiment features along with their relative importance

    • Ensemble information

    • An experiment preview

    • Word version of an auto-generated report for the experiment

    • A target transformations tuning leaderboard

    • A tuning leaderboard

  • Download AutoDoc: A Word version of an auto-generated report for the experiment. This file is also available in the Experiment Summary zip file. Note that this option is not available for deprecated models. Refer to the Automated Model Documentation (AutoDoc) section for more information.

Note: The “Download” options above (with the exception of AutoDoc) will appear as “Export” options if artifacts were enabled when Driverless AI was started. Refer to Exporting Artifacts for more information.

Experiment Insight and Scores

While an experiment is running, the UI provides you with options for viewing model insights (for time-series experiments) and model scores. Refer to Model Insights and Model Scores for more information.

Scores link