NSAP Eligibility Prediction

Overview

This project implements a machine learning solution to predict eligibility for the National Social Assistance Program (NSAP) in India. It uses demographic and socio-economic data to classify applicants into appropriate NSAP schemes, automating and improving the allocation process.

Problem Statement

The National Social Assistance Program (NSAP) provides financial assistance to the elderly, widows, and persons with disabilities from below-poverty-line (BPL) households through various sub-schemes. Manual verification and scheme assignment is time-consuming and error-prone. This project creates a multi-class classification model to streamline this process, ensuring timely and accurate benefit delivery.

Dataset

The project uses the district-wise pension data under NSAP available in:

• nsapallschemes.csv

Technology Stack

• IBM Watson Studio/Cloud Pak for Data: Primary platform for model development
• Watson Machine Learning (WML): For model deployment and serving
• AutoML: Used for automated model selection and optimization
• Python: Core programming language
• Jupyter Notebooks: For exploratory data analysis and model development

Project Structure

NSAP-Eligibility-Prediction/
├── assettypes/
│   ├── auto_ml.json                # AutoML configuration
│   └── wx_prompt.json              # Watson prompt configuration
├── assets/
│   ├── .METADATA/                  # Project metadata
│   ├── data_asset/
│   │   └── nsapallschemes.csv      # NSAP dataset
│   ├── environment/                # Python environment definitions
│   ├── notebook/                   # Jupyter notebooks for analysis
│   └── wml_model/                  # Watson Machine Learning models
└── README.md                       # This file

Implementation Details

The project follows these main steps:

1. Data Preparation: Loading and preprocessing the NSAP dataset
2. Exploratory Data Analysis: Understanding data distributions and relationships
3. Feature Engineering: Creating relevant features for better prediction
4. Model Training: Using IBM AutoML to select and optimize classification models
5. Model Evaluation: Assessing model performance with appropriate metrics
6. Deployment: Deploying the best model to IBM Cloud for inference

Setup and Usage

IBM Cloud Account Setup

• Create an IBM Cloud account or use existing credentials
• Set up Watson Studio/Cloud Pak for Data service

Project Setup

• Clone this repository
• Import the project into Watson Studio or run locally

Running the Project

• Execute notebooks in the recommended order
• Follow documentation within each notebook for specific instructions

Model Deployment

The trained model is deployed using Watson Machine Learning, allowing for:

• Real-time predictions via API
• Batch scoring for large applicant datasets
• Integration with government systems

Results

The model achieves [metrics to be added after evaluation] in predicting the appropriate NSAP scheme for applicants, potentially improving the efficiency and accuracy of benefit allocation.