📊 HR Employee Attrition Prediction using Logistic Regression

📌 Overview

This project focuses on predicting employee attrition (Yes/No) using Logistic Regression in Orange. The workflow includes data preprocessing, model building, and performance evaluation using ROC Curve and Confusion Matrix.

---

🎯 Objective

• Predict whether an employee will leave the company
• Identify key factors affecting attrition
• Evaluate model performance using classification metrics

---

🛠️ Tools & Technologies

• Orange
• Logistic Regression
• CSV Dataset

---

🔄 Workflow

1. Data Import (CSV File)
2. Feature Selection
3. Data Sampling (Train/Test Split)
4. Logistic Regression Model
5. Predictions
6. Evaluation (ROC Curve & Confusion Matrix)

---

📊 Results

• Accuracy: 84.3%
• AUC Score: 0.715
• Precision: 0.775
• Recall: 0.843
• F1 Score: 0.775

---

📈 Key Insights

• Job Satisfaction, Work-Life Balance, and Environment Satisfaction significantly impact attrition
• Model performs well for non-attrition cases
• Lower performance for predicting actual attrition (class imbalance issue)

---

⚠️ Limitations

• Model shows bias toward majority class (No Attrition)
• Nagelkerke R² = 0.11 (low explanatory power)

---

📚 What I have learnt

• Built and evaluated a Logistic Regression model end-to-end
• Learned key metrics, ROC-AUC, and handling class imbalance

---

📂 Project Files

https://1drv.ms/x/c/f483042b9735aab9/IQBuTG1kG7yUTowkiHjBAhLtAUGpzdPfJU2GjwXlHpw21lM?e=vYowkh → Input dataset

Orange work flow

https://1drv.ms/u/c/f483042b9735aab9/IQB7LGdyzxg7QqFAJA_OT81PAVsU9RcF26VuCRGbM1UzhfM?e=xjcKZ9

Model outputs

ROC Curve

Confusion Matrix

---

👩‍💻 Author - Dhathri Narne

---

⭐ If you like this project, give it a star!