Employee Attrition Risk Assessment

Explainable Machine Learning for HR Decision-Making

---

The problem

Every time a company loses an employee, it costs 50–200% of their annual salary to replace them. HR teams often find out too late — after resignation, not before.

Most attrition tools give a binary answer: will leave or won't leave.
That's not useful. HR needs to know who is at risk, how much risk, and exactly why.

---

What this system does differently

This is not a classifier. It's a decision-support system.

• Outputs a probability score (0–100% risk), not just yes/no
• Explains why each employee is flagged — using SHAP feature importance
• Lets HR simulate interventions — "what happens if we give a 10% raise?"
• Processes entire teams via CSV batch upload
• Generates downloadable prediction reports for HR review

---

Example output

Employee: John D. Attrition Risk: 78% ⚠️ High Top risk factors: ↑ Overtime hours (+23% risk contribution) ↑ Years since promotion (+18% risk contribution) ↓ Job satisfaction (+15% risk contribution) What-If: +10% salary increase → Risk drops to 61%

---

Features

Feature	What it enables
Probability risk score	Flexible thresholds — HR sets their own alert level
SHAP explainability	Every prediction justified, not just flagged
What-If salary simulation	Test interventions before implementing them
Batch CSV prediction	Score entire departments at once
Downloadable report	Ready for HR review meetings

---

Model

• Algorithm: Random Forest Classifier
• Output: Attrition probability (0.0 – 1.0)
• Features: Overtime, job satisfaction, years since promotion, salary band, tenure, and more
• Explainability: SHAP values per prediction
• Deployment: Lightweight inference pipeline — training and deployment separated for speed

---

Stack

Python scikit-learn SHAP Streamlit Pandas NumPy Joblib

---

Run locally

git clone https://github.com/AkashMs24/Employee-Attrition-Risk-Assessment-Using-Explainable-Machine-Learning.git
cd Employee-Attrition-Risk-Assessment-Using-Explainable-Machine-Learning
pip install -r requirements.txt
streamlit run app.py

---

Design decisions worth noting

• Probability over binary — a 78% risk score is actionable; "will leave" is not
• SHAP over black-box — HR can challenge and verify every recommendation
• Simulation over prediction — what-if analysis turns insight into intervention
• Separated pipelines — model trained offline, deployed artifact is lightweight for fast inference

---

Related projects

• Decision Intelligence System — ML + LLM business intelligence platform
• Fraud Detection System — XGBoost + SHAP + FastAPI
• FarmVoice AI — NLP + SHAP crop advisory

---

Built by Akash M S · Presidency University, Bengaluru
LinkedIn · GitHub · ms29akash@gmail.com