This project is part of an open-access BCG Virtual Experience Program with Forage.
BCG's client for this project is PowerCo, a major utilities company. PowerCo has had declining profits due to significant customer churn. BCG has been engaged to drive churn reduction within their Small & Medium Enterprise (SME) customers. As a data scientist, my task is to build a predictive model that can identify customers at high risk of churn.
- Business Understanding & Hypothesis Framing
-
understand business problem
-
formulate the hypothesis as a data science problem
-
lay out the major steps needed to test this hypothesis
-
communicate your thoughts and findings in an email
The final email is in the file
documents/Business Understanding & Hypothesis Framing.pdf
.
- Exploratory Data Analysis
-
perform exploratory data analysis (EDA) on the data
-
verify the hypothesis of price sensitivity being to some extent correlated with churn
-
provide a summary of findings and suggest next steps
The EDA is in the file
notebooks/eda.ipynb
.
- Feature Engineering & Modelling
-
clean and engineer features for prediction
-
predict churn probability with decision tree models
-
evaluate model performance
The corresponding notebook is in the file
notebooks/feature_engineering_and_modeling.ipynb
.
- Findings & Recommendations
-
make an executive summary of findings and provide recommendations
The final report is in the file
documents/executive_summary.pdf
.