Jade
This project analyzed customer churn at Jade Inc., an e-commerce company, using advanced machine learning techniques. By identifying key behavioral and transactional patterns, the project uncovered drivers of churn and provided targeted recommendations to improve customer retention and user experience.
Domain
Python, ML
The Goal:
The goal was to accurately predict customer churn and uncover the key factors driving attrition at Jade Inc., helping the business reduce churn, enhance customer satisfaction, and sustain growth. Through machine learning and deep exploratory analysis, the project aimed to deliver actionable intelligence that aligns with business strategy.
1
The Challenge:
The challenge involved cleaning and preparing complex data, dealing with missing values, outliers, and skewed distributions. With a wide variety of behavioral and survey-based features, selecting the right model to balance predictive performance and interpretability was critical. Designing recommendations based on model output—especially for UI and discount systems—required aligning technical results with human-centric experience data.
2
The Result
The Random Forest model achieved the best performance, with 91% accuracy and a 99% precision for predicting non-churners. Features like tenure and cashback amount were key contributors to churn. Actionable insights led to two major recommendations: simplifying the user interface and amplifying discount communication. These findings provided a strong foundation for retention strategies and future model development.
3
