CAP4001 Capstone Project | Vellore Institute of Technology-AP

Project work carried out during the penultimate semester of study for the credits prescribed under University Core of the curriculum, related to the specialization of the programme by applying the knowledge gained in the courses we have undergone so far.

Prediction of Health Insurance Amount Using Deep Neural Network

About The Project

The purpose of this project work was to build an efficient model to predict the health insurance amount. Various machine learning models were compared with Deep Neural Network model. With the evaluation metric being Mean Squared Error, DNN model outperformed the machine learning models, which is then used to serve the end users. The end-to-end application ensures the continuous integration and continuous deployment of the application to the users. Using Agile Software Development Lifecycle, the project is built following all the engineering standards, rules and regulations.

Built with

Flask, JavaScript, Azure, Docker,Visual Studio Code and Jupyter

Project set-up

Note for Windows user : Before step-3, uncomment line no.95 to 101 and delete the line no.116.

1. Clone the repository. sh git clone https://github.com/SnehaVeerakumar/Predict-health-insurance-amount.git
2. Activate virtual environment. sh virtualenv venv venv\Scripts\activate
3. Install necessary python packages. sh pip install -r requirements.txt
4. Start the server sh python app.py

Project set-up in Docker Desktop

1. Build using Docker Image sh docker build -t app_name:tag_name .
2. Run on port 5000 sh docker run -p 5000:5000 app_name:tag_name

Docker Hub Repository

Link : https://hub.docker.com/r/snehaveerakumar/insuranceprediction
Pull Image sh docker pull snehaveerakumar/insuranceprediction

Project folder description

1. Parent folder contains 3 important files.

   • app.py : Acts as the entry point to the application. It contains the URL routes and the codes for machine learning models.
   • Dockerfile : Contains the instructions to build a docker image.
   • requirements.txt : Contains the packages required to deploy the application.

2. Folder : Code

   Three sub folders in this folder contains the code for frontend and backend of the application. Jupyter Notebook has been used to analyse various machine learning models and deep neural network models and compared to select the best model using Mean Squared Error.

3. Folder : Dataset

   Contains all the dataset that is required for the project. Processed data and user data will also get saved in this folder for the further processing

Contributing

1. Fork the Project
2. Create your Feature Branch (git checkout -b test)
3. Commit your Changes (git commit -m 'message')
4. Push to the Branch (git push origin test)
5. Open a Pull Request

Dataset links

1. Disease indicators : https://www.kaggle.com/datasets/cdc/behavioral-risk-factor-surveillance-system
2. Insurance amount(insurance.csv) : https://www.kaggle.com/datasets/annetxu/health-insurance-cost-prediction

(back to top)