US Health Insurance Dataset

This project involves cleaning and preprocessing the US Health Insurance Dataset, training a machine learning model, and deploying it as an API using FastAPI. The API allows users to make predictions based on input data.

Table of Contents

• Dataset Identification
  • About this File
• Project Overview
• Built With
• Installation
  • Prerequisites
  • Cloning the Repository
  • Creating a Virtual Environment
  • Install Dependencies
• Running the API Locally
• Running the API with Docker
  • Build the Docker Image
  • Run the Docker Container
• API Endpoints
  • Home Route
  • Prediction Route
• File Structure
• Future Enhancements
• Contact
• License

Dataset Identification

About this File

This dataset contains 1,338 rows of insured data, where insurance charges are given against the following attributes: Age, Sex, BMI, Number of Children, Smoker, and Region. The attributes consist of both numerical and categorical variables.

Link to the dataset: Kaggle Dataset

Feature Name	Data Type	Description
Age	Integer	Age of primary beneficiary
Sex	String	Gender of the insurance contractor (male/female)
BMI	Float	Body mass index, indicating weight category
Children	Integer	Number of children covered by health insurance
Smoker	String	Whether the individual is a smoker (yes/no)
Region	String	Residential area in the U.S. (northeast, southeast, southwest, northwest)
Charges	Float	Individual medical costs billed by health insurance

Project Overview

• Data Cleaning & Preprocessing: Handled in Data Cleaning & Preprocessing.
• Model Training: The best model is selected and saved as a pickle file (best model).
• FastAPI Implementation: The API is built using FastAPI (main.py).
• Schema Definition: Schema.py defines input data validation using Pydantic.
• Deployment: The API is containerized for local deployment using Docker (Dockerfile).

Built With

• Python – Programming language.
• NumPy – Numerical computing.
• SciPy – Scientific computing algorithms.
• Pandas – Data manipulation and analysis.
• scikit-learn – Machine learning algorithms.
• Matplotlib / Seaborn – Data visualization.
• FastAPI – Web framework for building APIs in Python.

Installation

To run the API locally, follow these steps:

Prerequisites

Ensure you have the following installed:

• Python (>=3.8)
• FastAPI
• Uvicorn
• Pandas, NumPy, Scikit-learn
• Docker (for containerized deployment)

Cloning the Repository

git clone https://github.com/Just-Aymz/US-Insurance-Data.git
cd US-Insurance-Data

Creating a Virtual Environment

python -m venv venv
source venv/bin/activate  # On macOS/Linux
venv\Scripts\activate  # On Windows

Install Dependencies

pip install -r requirements.txt

Running the API Locally

uvicorn API_Files.main:app --reload

API available at: http://127.0.0.1:8000

Running the API with Docker

Build the Docker Image

docker build -t ml-api .

Run the Docker Container

docker run -p 8000:8000 ml-api

API Endpoints

Home Route

• URL: /
• Method: GET
• Response:

  {"message": "Welcome to the Machine Learning API"}

Prediction Route

• URL: /predict
• Method: POST
• Request Body:

  {
    "age": 25,
    "bmi": 24.5,
    "children": 2,
    "sex": "male",
    "smoker": "no",
    "region": "southwest"
  }

• Response:

  {"prediction": 1234.56}

File Structure

├── API_Files
│   ├── __pycache__/
│   ├── .gitkeep
│   ├── main.py
│   └── Schema.py
├── Dataset
│   ├── .gitkeep
│   └── insurance.csv
├── DockerFiles
│   ├── .dockerignore
│   ├── .gitkeep
│   └── Dockerfile
├── Preprocessing
│   ├── .gitkeep
│   └── Data_Preprocessing.ipynb
├── Static
│   ├── .gitkeep
│   ├── risk_management.jpg
│   └── style.css
├── Templates
│   ├── .gitkeep
│   ├── form.html
│   └── home.html
├── best_model_and_preprocessor.pkl
├── LICENSE
├── README.md
└── requirements.txt

Future Enhancements

• CI/CD pipeline for automated deployment
• Cloud deployment options (AWS, GCP, DigitalOcean)
• Expanding input validation and error handling

Contact

For questions or collaborations, please reach out via:

• Email: [email protected]
• LinkedIn: Amogelang More

License

This project is open-source under the MIT License.