Building and Comparing Predictive Models Using JMP ® Pro

Presenter: Andrea Coombs

Following Suggested Model Deployment Workflow

See how to:

  • Understand JMP Pro predictive modeling capabilities and basic predictive model types
  • Understand suggested workflow and which model to use at each step
  • Learn about the Diabetes Disease Progression case study data
  • Follow a modeling workflow that describes suggested models to use at each step
  • Use cross-validation to prevent overfitting candidate models and to help compare multiple candidate models
  • Use variable selection to determine variables to include in prediction model
  • Build Generalized Regression model using Adaptive Lasso estimation method
  • Build multiple candidate prediction models (Bootstrap Forest and Boosted Tree) that include risk factors identified during variable selection
    • Create validation column
    • Assess each model by comparing training and validation set results
    • Compare multiple models, including test set results, using R-squared, RMSE, Actual-by-Predicted, and Model Averaging
  • Deploy chosen model
  • Save  chosen model score code to Formula Depot to optionally deploy outside JMP

Note: Q&A included at times 45:55, 46:51, 47:18, 47:56, 49:52, 50:25 and 50:49.


Back to Top