Bengalathon 2019 (Deep Claims)

Introduction

Problem Statement: How can insurers leverage on historical and real-time data to predict future claims more effectively?

• We created a web-based portal which can be used to effectively access our ML model.
• The customer details can be entered using the 'Add' option and the customer details are saved to our Database.
• The customer is then evaluated by our ML model and is assigned a credibility score and classified as 'Credible' or 'Not Credible'.
• Previous Customers can also be seen and Customer details an be modified using the 'Edit' option.
• The ML model can easily be retrained once new customer data is obtained.

Machine Learning Model

Data pre processing

The data can be downloaded from here

The Data obtained had:

• Missing values
• A mix of continuous and categorical features
• Large disparities in the range of features

We tried a variety of techniques to combat these shortcomings:

• Regression to fill in continuous missing values
• Random Forest Classifiers to fill in categorical missing values
• One-hot encoding for the categorical features
• Logarithmic/Adaptive binning to convert the continuous features into categorical features
• Feature Scaling (min-max)
• Auto-encoders to get a small amount of meaningful parameters from the original Dataset

ML Architecture

We found that an ensemble of ML models outperformed the rest of the models.

It was observed that some customers were harder to classify than others. Thus, we made a semi-automatic model which passed such hard customers to the experts (Human Intervention). This led to a significant increase in accuracy.

Website

• Login Portal

• Home Screen

• Adding a new Customer

• Customer Details

Installing

• Setting up the virtual environment

cd Bengalathon
python3 -m venv env
source env/bin/activate
pip install -r requirements.txt

• Migrating the Database

cd portal
python manage.py makemigrations
python manage.py migrate
python manage.py createsuperuser

• Run Server

python manage.py runserver

• Demonstration
  • Dummy Company
    • Username: amul
    • Password: password

License

GNU GENERAL PUBLIC LICENSE

Version 3, 29 June 2007

Copyright (C) 2007 Free Software Foundation, Inc. https://fsf.org/

Everyone is permitted to copy and distribute verbatim copies

of this license document, but changing it is not allowed.