Predictive Modeling

After attending this course, you will be able to build, evaluate, interpret, improve and deploy both decision tree and scorecard models.

Duration: 1 day

Audience: Business and marketing analysts who want to use historical customer data to build and deploy predictive models

Prerequisites

Prior attendance at Segmentation and Profiling.

Topics covered

• Predictive modeling fundamentals
• What is a predictive model?
• Building and applying models
• Goals, populations and timelines
• Decision trees as a predictive modeling technique
• Model quality
• Measuring predictive power: R-square and Gini
• Ways to improve model quality
• Overfitting: what it is, how to measure it, and how to avoid it
• Scorecards
• What is a scorecard?
• Building and interpreting a scorecard in Decisionhouse
• Ways to improve scorecards
• Optimized binning
• Differences and similarities
• Between scorecards and decision trees
• Between binary and continuous outcomes
• Using your models
• Combining models
• Model deployment
• Modeling tips and pitfalls

Back