Prediction

Overview

Welcome to our introductory course on customer churn prediction using machine learning. This course is designed for non‑technical people who want to understand how machine learning can be applied to real‑world business problems.

The main objective is to introduce customer churn prediction using machine learning classification techniques. Customer churn refers to the phenomenon where customers stop doing business with a company (cancelling a subscription, closing an account, or simply stopping purchases).

Predicting customer churn is crucial for businesses because:

1. Retention is cheaper than acquisition — it’s generally more cost‑effective to keep existing customers than to acquire new ones.
2. Targeted interventions — by identifying customers likely to churn, companies can take proactive measures to retain them.
3. Improved customer satisfaction — understanding why customers might leave helps improve service and product offerings.

Machine learning can help predict customer churn by using past data about customers who have churned and those who haven’t to train a model. This model can then predict which current customers are at risk of churning.

This type of problem is called classification in machine learning — a supervised learning task where the goal is to predict which category an item belongs to. In our case:

• The items are customers
• The categories are “will churn” and “will not churn”

For this course, we’ll use a classification algorithm called K‑Nearest Neighbors (KNN). In short, it:

1. Looks at the features of a customer we want to classify.
2. Finds the K most similar customers in our training data.
3. Predicts the new customer will do whatever the majority of those K neighbors did.

There are many other classification algorithms available (e.g., Logistic Regression, Decision Trees, Random Forests), but KNN is a great starting point for understanding the fundamentals.