Customer
Churn
Prediction
Presented by:
Gunika Malhotra
Overview
Customer churn occurs when a customer ends
their relationship with a company. In a competitive
market, retaining customers is crucial for sustained
growth. This project aims to build a machine
learning model that predicts the likelihood of churn
based on customer demographics, account details,
and service usage patterns. By identifying at-risk
customers early, businesses can take proactive
measures to improve retention and reduce revenue
loss.
Problem
Statement
The objective of this project is to develop a
classification model that predicts whether a
customer is likely to churn. By leveraging
demographic features (like gender, senior status,
and tenure) and service-related attributes (such as
internet and phone services, online security, etc.),
the model helps identify high-risk customers. This
enables the company to implement targeted
retention strategies, enhancing customer
satisfaction and reducing revenue loss.
Loaded & Cleaned Dataset
Removed customerID.
Project Converted TotalCharges to numeric and handled
missing values.
Overview
Encoded Churn as binary (1 = churned, 0 = not churned).
Feature Engineering
Applied get_dummies() for categorical variables.
Scaled numerical features (tenure, MonthlyCharges,
TotalCharges) using StandardScaler.
Modeling
Used a Decision Tree Classifier to predict customer churn.
Evaluated with metrics: Accuracy, Precision, Recall, F1 score
Visualisation
Generated a correlation heatmap for feature relationships.
Tenure Matters
·Customers with shorter tenure are more likely to churn.
·Loyalty builds over time; the first few months are critical
for retention.
High Monthly Charges Increase Risk
Insights ·Customers paying more per month are more prone to
churn, possibly due to price sensitivity or lack of
perceived value.
Contract Type is Critical
·Month-to-month contracts have significantly higher churn
rates compared to annual contracts.
·Longer commitments reduce churn risk.
Payment Method and Paperless Billing
·Customers using electronic payments or enrolled in
paperless billing may behave differently.
·Some payment types (like electronic checks) show higher
churn rates.
Short Tenure → High Churn Introduce welcome offers,
onboarding support, and early-stage engagement for new
customers.
High Charges → High Churn Offer discounts, loyalty
credits, or flexible plans to customers with high monthly
bills.
Month-to-Month Contracts → High Churn Promote long-
term contracts with added benefits (e.g., price locks, free Risky Payment Methods Identify high-churn payment
upgrades). methods and offer alternatives or add verification
reminders.
Recommendations Contact High-Risk Customers Use churn predictions
to proactively reach out with retention offers.
Build a Dashboard Monitor churn metrics by customer
segment, contract type, and payment method in real-
time.
Thank You
