5  Modeling

CRISP-DM Phase 4. Select and apply modeling techniques and calibrate their parameters. Often loops back to Data Preparation.

5.1 Objectives

• Select modeling techniques — Choose the algorithm(s) and record modeling assumptions.
• Generate test design — Define how model quality will be tested (e.g., train/validation/test split, cross-validation).
• Build models — Run the techniques, capturing parameter settings and the resulting models.
• Assess models — Evaluate technical performance and rank models; revise parameter settings as needed.

5.2 Deliverables

• Modeling technique selection and assumptions
• Test design
• Models and parameter settings
• Model assessment

5.3 Notes

Document techniques, hyperparameters, experiments, and technical metrics here.