Can machine learning help predict and reduce employee turnover? Real-world HR modeling techniques uncover attrition patterns at a fictional automaker.
- Project Overview
- Features
- Tools & Technologies
- Usage
- Gallery
- Certificate
- References
- License
- Acknowledgements
- Author
Salifort Motors is a fictional car company facing high employee turnover. This project models and explains the drivers of attrition using a structured ML workflow:
- Data wrangling & preprocessing
- Exploratory data analysis (EDA)
- Predictive modeling with multiple classifiers
- Model evaluation and interpretability (SHAP, feature importance)
- Final recommendations for HR strategy and retention
Hosted online as an interactive web report aimed at both technical and general audiences.
- 📊 Interactive visual EDA (Seaborn, Matplotlib)
- 🤖 Four predictive models: Logistic Regression, Decision Tree, Random Forest, XGBoost
- 🔍 Model evaluation: confusion matrices, recall scores, misclassification analysis
- 🧠 SHAP and feature importances for explainability
- 💬 Executive summary with actionable business takeaways
- Language: Python
- Libraries: pandas, seaborn, matplotlib, scikit-learn, xgboost, statsmodels, shap
- Environment: Jupyter Notebook
- Deployment: GitHub Pages (HTML report)
All analysis can be found online at project site.
- Clone the repository.
- Install required dependencies from
requirements.txt. - Open the notebooks in the
notebooks/directory to explore the analysis:eda.ipynbfor exploratory data analysismodels.ipynbfor model development and evaluationexecutive_summary.ipynbfor a project overview and key findings
EDA Insights:
Model Results:
Final capstone project for Google Advanced Data Analytics Professional Certificate:
MIT License © 2025 Bryan Johns. See LICENSE for details.
- Thanks to Google and Coursera for the capstone structure.
- Style inspiration for header from Start Bootstrap Clean Blog
Bryan Johns, June 2025
bryan.johns.official@gmail.com | LinkedIn | GitHub | Portfolio