Real Estate Insurance Risk Calculator AI

Inspiration

California homeowners have been struggling with drought, and increasing amounts of wildfires. To make the situation worse home insurance providers are leaving the state as it has become impossible to charge a fair price for insurance premiums and be a sustainable business in California. My goal with this project was to help insurance providers identify which specific neighborhoods are most at risk of wildfire and potential property damage.

What it does

My project is a machine learning algorithm that can compute the risk level (high, medium, low) of a neighborhood. This kind of insight would reduce the cost of risk calculation at insurance companies allowing them to lower their premiums and provide better service to the homeowners of California.

How we built it

After settling on the idea for the project, I researched potential datasets and inputs that would be important for a fire risk model. I identified that wildfire historical data, population density, drought index, weather patterns, and topography would be a good start. For each of the datasets I isolated 1-3 key fields. Some of the key inputs I settled on were, latitude/longitude of closest wildfire, year of closest wildfire, amount of nearby fires historically (CA wildfire historical perimeters), elevation (NASA), population density (Census), total drought risk score (CA gov water shortage risk), and vegetation types. Once, I collected all the data I formatted it all into one spreadsheet file using pandas/geopandas libraries. Using Scikitlearn, I create a basic supervised machine learning algorithm to classify if a neighborhood (coordinates within a county), was high or low risk for wildfire.

Challenges we ran into

Testing, was definitely a big struggle with this project. The limited time I had to write the program and collect training datasets left me with very little time to do thorough and rigorous testing of the model.

Accomplishments that we're proud of

I am proud that my project truly has potential to be innovative and improve the lives of people. California residents are struggling with high premiums and insurance companies leaving the state. This puts residents at extreme risk especially during massive wildfire like what is happening in LA. My project aims to help mitigate risk and better the service provided by insurance companies which hopefully benefits the consumers as well.

What's next for Real Estate Insurance Risk Calculator AI

Though, I wish I could say the project is completed. It is not, but I plan to continue working on the project and explore better machine learning training algorithms and improved datasets. I believe there is a lot of value to be had out of this kind of software and plan to see it through.

Built With

• geopandas
• pandas
• python
• us-census-bureau