Inspiration

Our team thought it was a long, tedious, and tough process to spend time discovering a home that is right for you from it's cover all the way to the benefits and consequences it has for a person's future. We wanted to mitigate the effort in finding a perfect home that fits a person's profile and does not give someone a hard time in the long run.

What it does

This web application is developed in the Wix environment and utilizes machine learning to determine a top 5 recommendation list of houses that are suggested from a Wix data collection based on a user's inputs of income, taxes and insurance, monthly expenses, and age. The data processing that goes into the suggestions is determined with linear regression by a trained model that the website uses. The model was trained through a formula-based dataset that trained the model to make decisions of its own. We also started a Google Cloud Dialogflow integration that takes in the input parameters we ask the user to enter through speech. Ultimately, this would make it easier for users to provide their info to get home suggestions.

How we built it

We created our trained model for providing home suggestions through Python and hosted it on a Flask server that was put in the Heroku cloud. The Wix web application interface makes calls through a GET request to the server on the cloud for the home suggestions. The Wix web application interface has back-end Javascript that provides the key functionalities of the Wix web application.

Challenges we ran into

A challenge we ran into was having trouble making REST API requests to our python flask server on Heroku. We also continually faced small roadblocks making the exact functionality we wanted for our application through Wix.

Accomplishments that we're proud of

We're proud of the application's intelligence in making decisions about the price of a house. It took time to come up with an initial model with formulas to produce data that could be used effectively to make a trained model that powers our web application.

What we learned

We learned a lot about how to do UI design, code web functionality, and make REST API requests through the Wix environment. There was a lot of learning done to make the application what it is now, and we are grateful to the Wix developers' continuous and passionate support throughout the hackathon.

What's next for HouseMatch

HouseMatch needs to more time to foster an even better model that can be trained to make home recommendations that are far smarter and more visionary than a human's decision capabilities, so the machine learning aspect could definitely be assessed further. We could include more user inputs to better suit a user's needs such as location preferences. The application could also fully integrate our start of the Google Cloud Dialogflow integration in a mobile interface so that people could ask for home recommendations easily through speech.

Built With

Share this project:
×

Updates