🚧 PMGSY Scheme Classification using Machine Learning

This project automates the classification of rural infrastructure projects under different PMGSY schemes (PMGSY-I, PMGSY-II, PMGSY-III, RCPLWEA) using machine learning, based on their physical and financial characteristics.

---

📌 Problem Statement

Manual classification of road and bridge projects under PMGSY is slow, error-prone, and not scalable. Each scheme has different rules and objectives. This project uses machine learning to predict the correct PMGSY scheme, improving monitoring and planning.

---

📊 Dataset

• Source: AI Kosh - PMGSY Dataset
• Features: State, District, Road Length, Bridges, Project Cost, Completion, etc.
• Target: PMGSY Scheme (categorical)

---

🧠 Machine Learning Approach

• Model Used: XGBoost Classifier
• Platform: IBM Watson Studio (Cloud-based training)
• Deployment: IBM Watson Machine Learning (WML)
• Preprocessing:
  • Missing value handling
  • Label encoding & one-hot encoding
  • Train-test split

---

✅ Results

• Accuracy: 92.8%
• Macro F1-Score: 0.91
• Observations:
  • High accuracy for major schemes (PMGSY-I, II)
  • Lower performance on RCPLWEA due to class imbalance

---

🧰 Tools & Technologies

• IBM Cloud Lite
• IBM Watson Studio
• IBM Watson Machine Learning
• Python (pandas, numpy, scikit-learn, xgboost, matplotlib)

---

🚀 Future Improvements

• Balance class distribution for better RCPLWEA accuracy
• Integrate real-time PMGSY project data using APIs
• Expand dataset with more districts and years

---

🙋‍♂️ Author

K Adithya
CMR University – CSE Data Science