ML in Insurance

• 

  Decision Tree used for classification

• 

  Example Form used to capture user data on which policy recommendation is output

Inspiration

After a recent troublesome experience choosing optimal insurance policies experienced by a team member, we were curious to find a way to make insurance recommendations more data-driven instead of hearsay.

What it does

Our ML model gives real-time recommendations to insurance agencies and agents which allows them to rely on past data and parameters to suggest the best health insurance tier to customers.

How we built it

We used scikit-learn framework to implement the ML model and fed it past user data with policy tiers. We split the 50 rows of customer data we have used into training and testing with a split of 80-20. We implemented a Decision Tree algorithm to classify the data.

Challenges we ran into

Insufficient time to build an external API (Used Google Forms internal API instead). We would like to continue work on this project by building a node.js API and a functional back-end.

Accomplishments that we're proud of

Building a functional ML model with decent accuracy despite constraints of limited data

What we learned

Learning about the different types of ML models, frameworks and algorithms and choosing the optimal one for our use case.

What's next for ML in Insurance

More specific recommendations with alternative options and ideas for the client could be implemented with more data and parameters. Making insurance more holistic instead of being only catered to health, life or home individually.

Built With

• colab
• python
• scikit-learn