CRISP-DM (Cross-Industry Standard Process Data Mining)

CRISP-DM is a structured approach to data mining projects with six phases: Business Understanding, Data Understanding, Data Preparation, Modelling, Evaluation, and Deployment.

Goals

Define project objectives and requirements Collect, describe, and explore data Clean, transform and integrate data Select, create, and validate models Evaluate model performance Plan and monitor the deployment, maintenance, and monitoring of the model.

Best pratices

Clearly define project objectives and requirements in the Business Understanding phase Invest adequate time in the Data Understanding and Preparation phases to ensure quality data Evaluate multiple models and select the best-performing one Continuously monitor and maintain the deployed model.

Related content