Customer Churn Analysis

Built by Sahil Narula. This project turns raw customer data into a clear business story about why customers leave, what matters most, and how the analysis was done.

View GitHub Repo See the Charts
48charts presented clearly
10+models compared

Project summary

Business problem

Understand which customer behaviors are linked to churn so the business can take action earlier.

What I did

I cleaned the dataset, explored the main patterns, built models, compared results, and tuned the best ones.

What I got

A clear churn analysis, a set of model results, and a website that makes the work easy to present.

How I worked

01 Clean and prepare the data
02 Study customer behavior
03 Compare machine learning models
04 Tune the strongest models

Tools I used

Python Pandas NumPy Scikit-learn XGBoost CatBoost Plotly Matplotlib Seaborn

Key Insights

Monthly plans, billing habits, service usage, and shorter tenure are strongly linked to churn risk.

Full visual story Every chart from the notebook is shown below with a simple title.

How To Reduce Customer Churn

Improve Early Customer Experience

Focus on the first 90 days with onboarding support, proactive check-ins, and setup guidance for new users.

Offer Better Contract Incentives

Encourage long-term plans with loyalty discounts and clear value communication for month-to-month customers.

Target High-Risk Segments

Use churn signals such as billing behavior and support usage to trigger retention offers before cancellation.