Skibidi Insurance Application

Overview

The Skibidi Insurance Application is a multi-platform solution designed to enhance user experiences in choosing and managing insurance policies. This application utilizes Machine Learning models to suggest the most suitable insurance policies based on user-inputted lifestyle data. The architecture is robust, featuring a 3-tier model with a frontend built in Flutter, a Django backend API, and a MySQL database managed on Microsoft Azure.

Features

• User Authentication: Secure email-based user authentication system.
• User Dashboard: Displays user profile and insurance policy details.
• Suggested Policies: Utilizes ML models to suggest personalized insurance policies based on user lifestyle data.
• Policy Management: Users can view, purchase, and manage their insurance policies.
• Multi-Platform Support: Available on iOS, Android, Web, macOS, and Windows.

Business Use Cases

1. Policy Recommendation: Users receive insurance policy suggestions based on their lifestyle data.
2. Policy Retrieval: Users can retrieve and view details of their purchased policies.
3. Policy Purchase: Users can buy recommended insurance policies directly through the app.

Application Architecture

Skibidi follows a 3-tier architecture:

• Presentation Layer: Implemented using Flutter to provide a cross-platform user interface.
• Application Layer (Business Logic): Built with Django, this layer handles the core functionality and business logic of the application.
• Data Layer (Database): Uses MySQL for data storage and management, initially set up locally and migrated to Azure for scalability and security.

Application Architecture

Backend API

The backend is structured to handle various types of requests:

• Query data
• Modify/add data
• Delete information

The API routes include:

• POST /add_customer: Adds a new customer.
• POST /add_policy: Adds a new policy.
• POST /purchase_policy: Endpoint for purchasing a policy.
• GET /policies: Retrieves all policies.
• GET /policies/{type}: Fetches policies by type.
• GET /obesity_level: Determines the obesity level of the user.
• GET /purchased_policies: Fetches policies purchased by the user.

Database

The database uses MySQL with optimizations for fast data retrieval:

• Object Relational Mapping (ORM): Django ORM is used for database interactions.
• Indexing: Applied on frequently accessed tables to improve performance.
• Normalization: Ensures data integrity and reduces redundancy.

Updating Insurance Policy Quotes Using Machine Learning

Machine Learning is utilized to adjust insurance quotes based on user lifestyle data. The models include:

• K-Nearest Neighbors (KNN)
• Support Vector Classifier (SVC)
• Decision Tree Classifier
• Random Forest Classifier
• AdaBoost Classifier
• Gradient Boosting Classifier
• Stochastic Gradient Descent (SGD) Classifier

The ML workflow updates policy quotes and suggests relevant policies based on the user's predicted obesity level.

Version Control

The project uses GitHub for version control, organized under the Skibidi Insurance Project organization. The repositories are:

• Skibidi-be: Contains the backend API and ML models.
• Skibidi-Insurance: Contains the frontend application developed in Flutter.

Setup and Installation

To set up the Skibidi Insurance Application:

1. Clone the repositories:

   git clone https://github.com/Skibidi-Insurance-Project/Skibidi-be.git
   git clone https://github.com/Skibidi-Insurance-Project/Skibidi-Insurance.git
   

2. Install dependencies for the backend (Django):

   cd Skibidi-be
   pip install -r requirements.txt
   

3. Install dependencies for the frontend (Flutter):

   cd Skibidi-Insurance
   flutter pub get
   

4. Start the Django backend server:

   python manage.py runserver
   

5. Run the Flutter application:

   flutter run
   

License

This project is licensed under the MIT License - see the LICENSE file for details.

Authors

• Apoorv Singh
• Raghav Mantri

Acknowledgments

Special thanks to NYU and Dr. Jean-Claude Franchitti for their guidance and support throughout the development of this project. Thanks to all contributors and testers who helped in refining the application.