Machine learning starts with a precise decision and prediction time. Features must be available at that time, and the target must reflect the outcome you want to predict.

Worked example

Target: customer cancels within the next 30 days. Features: information available at the end of the current month.

Put it into practice

Write a churn problem statement naming the prediction time, target, and action that follows a high-risk prediction.

Check your understanding

• Explain the main idea in your own words.
• Compare your result with the worked example and identify any assumptions you made.
• Record one question or edge case to discuss with your instructor.

Connect this to your project

Build a customer churn classifier with a preprocessing pipeline, a baseline comparison, and a model evaluation report.

Keep the output of this activity in your project folder. You will use it when you review and present your final work.