Costco Bears

Inspiration

Artificial Intelligence is a booming industry, with skills such as machine learning and deep learning becoming extremely valuable. Companies all around the world are looking for ways to implement ML algorithms in their software to improve customer service or learn valuable insights. One such noble company, Fannie Mae, has had a rich, diverse industry in providing innovative housing solutions, with one potential gap being effective, accurate prediction of home sales. Our team sought not only to fix this problem, but to go above and beyond by integrating augmented reality and emphasizing an efficient, streamlined pattern for users to acquire and analyze home prices in their surrounding area with an accessible web application. In memory of George Tang.

What it does

Our platform emphasizes the importance of having a simple, yet effective method for completing tasks. Users can select a three digit zip code region and predict reliable home sales (origination dates of loan acquisitions) there for a given period in the future. Given the thousands of existing variables provided by Fannie Mae associated with individual loans and national economic indicators, machine learning was the most obvious choice.

How we built it

We used the Fannie Mae API to retrieve data, HTML, CSS, and JS to create the website application, and Keras and Numpy to create the machine learning backend for predicting home sale prices. We used most of our data to build the predictor and the remaining to verify that the predictions were correct.

Challenges we ran into

We needed to evaluate which indicators were useful and what loan information could be discarded to develop our predictive model. These human choices, on top of the difficulty of navigating an unfamiliar API, proved challenging.

Accomplishments that we're proud of

We are proud that we were successfully able to create a website that uses machine learning to make valuable predictions using economic and housing data provided by the Fannie Mae API.

What we learned

We learned a plethora of concepts, ranging from the integration of GET Requests with software and utilizing the Fannie Mae API to Data visualization software and integrating multiple website components. We also worked on highlighting the most important components of our application and presenting them effectively to an audience.

What's next for Costco Bears

In the future, we hope to expand our dataset by looking at different sources, such as crime data and working with other local companies to acquire more housing-related data to further mature our machine learning algorithm.