Inspiration
Our project was inspired by a careful look at the Homes.com website, in which our team noticed certain important things missing, especially within the lookup-houses form.
What it does
Our project aims to modify and enhance the user experience by providing them with more personalized housing recommendations based on the inputs they enter on the form. We go beyond geographical and financial factors to offer users an opportunity to look up places based on factors such as quality of education, public transport availability, water purity index, etc.
How we built it
We made a replica of Homes.com's Home Page, and attached a House Finder form to it. The form collects user inputs, and then in the back end, the user's information gets put through a linear regression model trained on a self-built data set. The final page shows a potential home that the user may be interested in buying.
Challenges we ran into
When deciding on the prediction model to use we debated using the K Nearest Neighbors model. This would have more easily allowed us to predict a batch of homes the user would be interested in. The problem became the data. The data was self-created and completely randomized therefore predictions weren't reasonable given the preferences of the user. To solve this we made a positive correlation between prices/budget and Home purchased and used a linear regression model to have the most accurate results.
Accomplishments that we're proud of
We are proud of the full end product. The user interface was well-designed and very intuitive. We are also proud of how quickly the team learned to use different tech stacks like Flask and JavaScript. Some members of our team had never worked on a project before and they contributed a lot.
What we learned
We learned the importance of planning. We planned agile sprints while working on this project and we ended up spending more time than we had planned on some tasks. We spent a lot of time cleaning and formatting the user data to be able to integrate the front end with the back end. We realized that we would have saved ourselves a lot of time if we had requested the user information in the same format that we needed to process it. This made us realize how important it is to communicate every step with your team and plan far ahead. I believe that we could have finished the project earlier if the front-end developers and the back-end developers had planned together more thoroughly. However, we completed our other tasks very quickly so we were still able to finish our project in time.
What's next for CoStar Group Home Finder Extension
If we had more time we would have implemented an extension that allows users to compare multiple recommended houses to each other. We also wanted to add information about local events and activities so that users can know what their social life will be like in their new home.
Log in or sign up for Devpost to join the conversation.