Model validation, a method of determining if a predictive model is generalizable to new data, is critical when using models to help make decisions. Data can be partitioned into sets before modeling to avoid overfitting. Part of the original data is used to estimate parameters and the rest of the data is used to tune or evaluate the parameters.

JMP Pro incorporates validation into some of its models and allows away to interactively create data partitions in others.

See how to:

• Understand the value and importance of validating your models.
• Choose random, time-based, or balanced validation data sets.
• Find ways to start building each of the four families of models.
• Understand how test, training and validation sets work.
• Use different JMP Pro modeling techniques.
• Screen models.
• Deploy models to put into action the insights they provide.

This webinar covers: validation techniques, decision trees, model scoring, and model selection.

Key to predictive modeling is using validation columns:

Resources

• Neural Network Tuning for JMP Pro JMP Extension from JMP Marketplace
• Torch Deep Learning for JMP® Pro JMP Extension from JMP Marketplace
• Model from demo on JMP Live
• Accelerating Innovation using Bayesian Optimization video and materials
• Imbalanced Classification JMP Extension from JMP Marketplace
• Documentation on predictive models using JMP Pro
• Documentation on model screening to fit and compare predictive models