MLS-C01 Question 125
Single answerA retail company wants to reduce the number of customer churns by identifying customers who are likely to cancel their subscription. They have historical customer data, including demographic information, subscription details, and customer behavior metrics. As a Machine Learning Specialist, how should you frame this business problem as a machine learning problem?
- A
Frame it as a regression problem to predict the exact number of days until a customer churns.
- B
Frame it as a classification problem to predict whether a customer is likely to churn or not.
- C
Frame it as a clustering problem to group similar customers based on their likelihood to churn.
- D
Frame it as a reinforcement learning problem to determine the optimal actions to prevent churn for each customer.
Show answer and explanation
Correct answer: B
Explanation
Customer churn prediction is best framed as a binary classification problem. The business goal is to identify customers who are likely to churn, which requires a supervised learning approach with labeled data (churn or not churn). Classification algorithms can then be used to predict the likelihood of churn for each customer.
- A. Incorrect.
Incorrect: Regression is used to predict continuous numerical values. Predicting whether a customer is likely to churn is a binary outcome (churn or not churn), making regression an inappropriate choice.
- B. Correct.
Correct: Customer churn prediction is a classic binary classification problem where the goal is to predict whether a customer will churn (yes or no) based on historical data.
- C. Incorrect.
Incorrect: Clustering is an unsupervised learning technique used for grouping data without predefined labels. While clustering can provide insights into customer segmentation, it does not directly solve the problem of predicting churn.
- D. Incorrect.
Incorrect: Reinforcement learning is used for sequential decision-making problems with rewards. Preventing churn could involve reinforcement learning in specific cases, but predicting churn itself is not a reinforcement learning problem.