Data Science Job Simulation – September 2025
This project simulates the role of a Junior Data Scientist at BCG X, working with the client PowerCo – a major gas & electricity utility.
The goal: analyze churn, build predictive models, and deliver business recommendations to improve customer retention.
📂 assets/
📄 certificate.png – Forage certificate
📂 notes/
📄 LEARNINGS.md – Reflections & key skills gained
📄 NOTES.md – Detailed workflow documentation
📂 task_1_business_understanding/
📄 task1.md – Business Understanding & Problem Framing
📂 task_2_eda/
📂 data/ – Raw & cleaned data files
📂 notebooks/ – Jupyter notebooks for EDA
📄 task2.md – Exploratory Data Analysis & Data Cleaning
📂 task_3_feature_engineering/
📂 data/ – Feature-engineered data
📂 notebooks/ – Notebooks for transformations
📄 task3.md – Feature Engineering summary
📂 task_4_modeling/
📂 data/ – Model-ready datasets
📂 notebooks/ – Model training & evaluation notebooks
📄 task4.md – Modeling results
📂 task_5_executive_summary/
📄 task5.md – Executive Summary deliverable
📄 intro.md – Introduction to BCG X & PowerCo
📄 README.md – Main landing page
- 🌐 Intro – BCG X & PowerCo Context
- Task 1 – Business Understanding
- Task 2 – EDA & Data Cleaning
- Task 3 – Feature Engineering
- Task 4 – Modeling
- Task 5 – Executive Summary
- 🎓 Learnings
- 📝 Notes
- Task 1 – Business Understanding → Framed churn as a price sensitivity problem & outlined data needs.
- Task 2 – EDA → Cleaned messy data, found churn rate ~10%, and identified sales channel & contract type as key drivers.
- Task 3 – Feature Engineering → Applied log transforms, one-hot encoding, and date-derived features for model readiness.
- Task 4 – Modeling → Built Decision Tree, and Random Forest; Random Forest achieved ~50% recall, the key business metric.
- Task 5 – Executive Summary → Consolidated findings into a client-facing recommendation for PowerCo.
Forage – BCG X Data Science Job Simulation (Sept 2025)
Hem Gandhi
✨ This repository demonstrates an end-to-end data science workflow — from business framing → EDA → feature engineering → modeling → executive recommendations — packaged in a recruiter-ready portfolio project.