# 💰 Pay Predict

AI-powered salary prediction tailored to the Indian tech market. Built with **Streamlit**, **scikit-learn** and an interactive dark-orange UI.

---

## 🚀 Live Demo
Run the app locally (instructions below) and open it in your browser; Streamlit starts on `localhost:8501` by default.

---

## ✨ Features
- **89 % R² accuracy** via Gradient Boosting Regressor.  
- **Real-time results** on every input change.  
- **Dark + orange theme** with mobile-friendly layout.  
- **Market benchmarks** & interactive Plotly charts (job, city, education).  

---

## 🏗️ Project Layout

pay-predict/ ├── app_streamlit.py # Streamlit front-end ├── train_model.py # Model training pipeline ├── requirements.txt # Python dependencies ├── model.joblib # Trained model (generated) ├── label_encoders.joblib # Encoders (generated) └── indian_salary_data_500.csv

---

## ⚙️ Tech Stack
| Layer            | Tools / Libraries |
|------------------|-------------------|
| User Interface   | Streamlit         |
| Machine Learning | scikit-learn (Gradient Boosting) |
| Data Handling    | pandas · numpy    |
| Visualisation    | Plotly            |
| Persistence      | joblib            |

---

## 🛠️ Quick Start

1. Clone repository

git clone cd pay-predict

2. Install dependencies

pip install -r requirements.txt

3. Add dataset

Place indian_salary_data_500.csv in the project root

4. Train the model (generates model.joblib & label_encoders.joblib)

python train_model.py

5. Run the application

streamlit run app_streamlit.py

---

## 📈 Dataset
Indian software-sector salary survey with  
• 500+ records • 8 input features (age, gender, education, experience, job title, location, city, nationality)  
• Target: annual salary in ₹ lakhs.

---

## 🤖 Model
GradientBoostingRegressor  
• 200 estimators • max_depth 6 • learning_rate 0.1  
Performance on held-out test set: **R² ≈ 0.89**.

---

## 📝 License
MIT © 2025 Pay Predict