Inspiration

The Drive Safe and Save program from StateFarm is one of a kind, fetching information regarding driver behavior patterns by means of a beacon. When we look at scaling this technology to include home and personal insurance, we already have a lot of data points to get this information from. In our case, the source of our data is an array of smartwatch and smartphone accelerometer and gyroscopic readings to judge the activity level of the person. Any insurance company, more than wanting less-risk clients, expects it's clients to be in the best of their health. What better way to incentivize clients than offering better discounts for better well-being. The IncentivizeR is win-win, for companies and customers!

What it does

The Index IncentivizeR calculates a close approximation of the person's BMI and hence, his health, using features based on his face. Based on the timestamp reading over a specific period of time, we can determine the client's incentive score. This is very similar to how the credit score works. Better customer spending habits and financial discipline, better are his chances of fetching loans and higher credit lines. The script logs the data in real-time to a Google cloud firebase database to log the data and perform statistics off it.

How I built it

We first, developed a Regression model to estimate the parameters which had the most significance in the incentivizer index. Based on this study, BMI was found to affect it the most. We then proceeded to develop a facial to BMI mapper to predict the BMI. Finally, we used an algorithm to combine the attributes according to the weights with which they affect the index to calculate the final index.

Challenges I ran into

1) Collecting our own sample data to train and test the model was challenging part as we collected our own photos and BMI including our friends and some we collected through the UCI repository and merged with our's to increase the sample size. 2) While recognizing the images there were a lot of noises in the background which make made predictions a little bit difficult.

Accomplishments that I'm proud of

1) Created a model to perform facial recognition 2) Calculated BMI index using facial adiposity 3) Created a system that calculates a health score ( similar to a credit score) 4) Hosted the website service on Google Cloud Platform using the Firebase service

What I learned

1) Implementing a facial recognition algorithm 2) Calculating BMI just by using facial features rather than using the conventional and tedious method.

What's next for Life Insurance Index IncentivizeR

1) Develop a highly scalable model using Google Api's to track the shopping and eating habits of the customer to offer incentives and discounts on places most visited.

2) Develop a model on a similar front for offering home and renter's insurance.

Built With

Share this project:

Updates