👥 HR Employee Attrition Analysis & Prediction

A data-driven HR analytics solution combining EDA, Random Forest ML, and an interactive Streamlit dashboard to help organizations identify and reduce employee attrition.

---

📌 Project Overview

This project analyzes employee attrition using HR data and builds insights into why employees leave an organization. It combines:

• 📊 Exploratory Data Analysis (EDA) — uncovering patterns across 31 features
• 🤖 Machine Learning — Random Forest classifier for attrition prediction
• 📈 Interactive Streamlit Dashboard — real-time KPIs, prediction, and visualizations

The goal is to help HR teams make data-driven decisions to improve employee retention.

---

🎯 Target Variable

Value	Meaning
Yes (1)	Employee left the company
No (0)	Employee stayed

---

📊 Dataset Summary

Property	Detail
Total Records	1,470 employees
Total Features	31 (after cleaning)
Missing Values	None
Data Type	Structured HR Dataset

Removed columns:

• Constant columns — Employee Count, Standard Hours
• Irrelevant derived features — CF columns

---

🧾 Feature Categories

👤 Demographic Features

• Age, Gender, Marital Status, Education, Education Field

💼 Job-Related Features

• Department, Job Role, Job Level, Business Travel, Over Time

💰 Compensation Features

• Monthly Income, Daily Rate, Hourly Rate, Monthly Rate, Percent Salary Hike, Stock Option Level

😊 Performance & Satisfaction

• Job Satisfaction, Environment Satisfaction, Relationship Satisfaction, Work Life Balance, Job Involvement, Performance Rating, Training Times Last Year

📅 Experience & Tenure

• Total Working Years, Years At Company, Years In Current Role, Years Since Last Promotion, Years With Current Manager, Num Companies Worked, Distance From Home

---

📈 Exploratory Data Analysis (EDA)

✔ Univariate Analysis

• Distribution of Age, Monthly Income, and Attrition rate

✔ Bivariate Analysis

• Monthly Income vs Attrition
• Years at Company vs Attrition
• Job Satisfaction vs Attrition
• Over Time vs Attrition

✔ Multivariate Analysis

• Correlation heatmap across all numerical features

---

🔍 Key Insights

Insight	Finding
💸 Income	Employees with lower income are more likely to leave
⏳ Tenure	Employees with shorter tenure show higher attrition
😓 Overtime	Overtime significantly increases attrition risk
😐 Satisfaction	Low job satisfaction strongly correlates with leaving
📊 Salary	Strongly linked with experience and job level
🔁 Tenure features	Highly correlated with each other

🎯 High-Risk Employee Profile

• Younger & less experienced employees
• Lower salary bracket
• Working overtime with low job satisfaction

---

🤖 Machine Learning Model

Property	Detail
Algorithm	Random Forest Classifier
Accuracy	85%+
Preprocessing	StandardScaler for feature scaling

Input Features used for prediction:

• Age
• Monthly Income
• Years at Company
• Job Satisfaction
• Over Time

---

🖥️ Streamlit Dashboard

The interactive dashboard includes:

• 📊 KPI Metrics — Overall attrition rate, average salary, average tenure
• 🔍 Data Exploration — Filter and explore employee data interactively
• 🤖 Attrition Prediction — Enter employee details and get instant risk prediction
• 📈 Feature Importance — Visualize which factors drive attrition the most

---

📸 Screenshots

Coming soon — upload screenshots to the screenshots/ folder and update the paths below

screenshots/
├── kpi_dashboard.png
├── eda_charts.png
├── prediction_panel.png
└── feature_importance.png

---

🛠️ Tech Stack

---

🚀 Getting Started

1. Clone the repository

git clone https://github.com/Shrihariniselvakumar/HR-Employee-Attrition-Analysis-Prediction.git
cd HR-Employee-Attrition-Analysis-Prediction

2. Install dependencies

pip install -r requirements.txt

3. Run the app

streamlit run app.py

---

📁 Project Structure

HR-Employee-Attrition-Analysis-Prediction/
│
├── app.py                      # Streamlit dashboard
├── model/
│   ├── train_model.py          # Random Forest training script
│   └── attrition_model.pkl     # Saved model
├── data/
│   └── hr_dataset.csv          # IBM HR Analytics dataset (1470 records)
├── notebooks/
│   └── EDA.ipynb               # Full EDA notebook
├── screenshots/                # App screenshots
├── requirements.txt
└── README.md

---

🔮 Future Improvements

• Add SHAP explainability for individual predictions
• Retention strategy recommendations per high-risk employee
• Connect to live HRMS data via API
• Deploy on Streamlit Cloud
• Email alerts for HR managers on high-risk employees

---

👩‍💻 Author

Shri Harini Selvakumar

---

📄 License

This project is open source and available under the MIT License.