🎯 Rank Predictor

Machine Learning web application that predicts a student's expected JEE Rank using historical exam-related data such as marks, percentile, year, and total number of candidates.

The project demonstrates a complete end-to-end Machine Learning workflow including data preprocessing, model training, evaluation, visualization, prediction export, and deployment using Streamlit.

---

📌 Overview

Rank Predictor uses regression-based Machine Learning models to estimate a student's probable JEE rank based on exam statistics.

The project was built to:

• understand regression algorithms
• Data visualization
• practice preprocessing pipelines
• implement real-world ML workflows
• deploy an ML model using Streamlit

The application currently uses:

• Linear Regression
• Random Forest Regressor

where Random Forest provides significantly better accuracy and prediction performance.

---

🚀 Features

✅ JEE Rank Prediction
✅ Machine Learning Regression Models
✅ Data Preprocessing Pipeline
✅ Feature Scaling & Missing Value Handling
✅ Data Visualization using Matplotlib
✅ CSV Export of Predictions
✅ Streamlit Interactive UI
✅ Real-time Rank Prediction
✅ Model Evaluation Metrics
✅ End-to-End ML Workflow

---

🛠️ Technologies Used

Technology	Purpose
Python	Programming Language
Pandas	Data Handling
NumPy	Numerical Operations
Matplotlib	Data Visualization
Scikit-learn	Machine Learning
Streamlit	Web Application UI

---

📂 Project Structure

Rank-Predictor/
│
├── app.py
├── model.py
├── Jee_data.xlsx
├── linear_regression_predictions.csv
├── random_forest_predictions.csv
├── requirements.txt
├── README.md
└── images/

---

⚙️ Machine Learning Workflow

Data Collection
        ↓
Data Cleaning
        ↓
Feature Engineering
        ↓
Train-Test Split
        ↓
Preprocessing Pipeline
        ↓
Model Training
        ↓
Model Evaluation
        ↓
Prediction Export
        ↓
Streamlit Deployment

---

📊 Dataset Features

The dataset includes:

Feature	Description
Year	Examination Year
Marks	Student Marks
Percentile	JEE Percentile
Total_Candidates	Total Students Appeared
Rank	Target Variable

---

📈 Data Visualization

The project includes multiple visualizations:

• Marks vs Percentile
• Marks vs Rank
• Percentile vs Rank
• Marks Distribution

These plots help understand:

• feature relationships
• rank trends
• score distribution
• dataset patterns

---

🤖 Machine Learning Models

1️⃣ Linear Regression

Linear Regression was used as a baseline regression model.

Advantages

• Simple
• Fast
• Easy to understand

Limitations

• Performs poorly on nonlinear relationships
• Can predict negative values

---

2️⃣ Random Forest Regressor

Random Forest Regressor is the primary model used in the project.

Advantages

• Handles nonlinear relationships
• Better prediction accuracy
• Reduces overfitting
• More stable predictions

Performance

Random Forest significantly outperformed Linear Regression in:

• RMSE
• MAE
• R² Score

---

📏 Evaluation Metrics

The models are evaluated using:

Metric	Description
RMSE	Root Mean Squared Error
MSE	Mean Squared Error
MAE	Mean Absolute Error
R² Score	Model Accuracy Score

---

🧹 Data Preprocessing

A complete preprocessing pipeline was implemented using:

• SimpleImputer
• StandardScaler
• ColumnTransformer
• Pipeline

This automates:

• missing value handling
• feature scaling
• preprocessing workflow

---

🌐 Streamlit Web App

The project includes an interactive Streamlit web application where users can:

✅ Select Exam Year
✅ Enter Marks
✅ Enter Percentile
✅ Enter Total Candidates
✅ Predict Expected JEE Rank

---

🖥️ Streamlit Preview

🎯 JEE Rank Predictor

Select Year: 2025
Enter Marks: 250
Enter Percentile: 99
Enter Total Candidates: 1400000

🎯 Predicted JEE Rank: 10366

---

▶️ Installation & Setup

1️⃣ Clone Repository

git clone https://github.com/your-username/Rank-Predictor.git

---

2️⃣ Navigate to Project Directory

cd Rank-Predictor

---

3️⃣ Install Dependencies

pip install -r requirements.txt

---

4️⃣ Run Streamlit Application

streamlit run app.py

---

📦 Generate Requirements File

pip freeze > requirements.txt

---

📤 Prediction Export

The project exports prediction results into CSV files:

• linear_regression_predictions.csv
• random_forest_predictions.csv

The exported files include:

• actual rank
• predicted rank
• prediction error

---

📈 Example Prediction

Year	Marks	Percentile	Total Candidates	Predicted Rank
2025	250	99	1400000	~10366

---

🔥 Future Improvements

• Hyperparameter Tuning using GridSearchCV
• XGBoost Integration
• CatBoost Regressor
• Deep Learning Models
• Cloud Deployment
• Live Rank Analysis
• User Authentication
• Advanced Analytics Dashboard
• Model Persistence using Joblib
• Docker Deployment

---

💡 Key Learning Outcomes

This project helped in learning:

✅ Regression Models
✅ Machine Learning Pipelines
✅ Data Visualization
✅ Feature Engineering
✅ Model Evaluation
✅ Streamlit Deployment
✅ End-to-End ML Workflow
✅ Real-world ML Project Structure

---

📚 Concepts Used

• Supervised Learning
• Regression
• Ensemble Learning
• Feature Scaling
• Data Preprocessing
• Model Evaluation
• Machine Learning Deployment

---

📜 License

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

---

👨‍💻 Author

Arpit Shirbhate

• Data Science Enthusiast
• Machine Learning Learner
• Open Source Contributor

---

⭐ Support

If you found this project useful:

⭐ Star the repository
🍴 Fork the project
🛠️ Contribute improvements
🚀 Share with others