HouseKachow

Our program matches you with houses by letting you compare two and refining the algorithm to suggest properties that fit your preferences and past interactions. At the end, the best houses are offered

Comment

Inspiration

Our project was inspired by the idea of creating a personalized and efficient way of matching individuals with their ideal homes. We wanted to provide a user-friendly platform that would allow users to interact with different properties and provide feedback on their preferences. By incorporating machine learning algorithms and data analysis techniques, we aimed to refine our recommendations based on users' past interactions and preferences. Our goal was to create a tool that could streamline the home buying process and make it more enjoyable for everyone involved.

What it does

The project we worked on is a house matching system that helps users find properties that match their preferences and past interactions. It presents users with two options at a time and learns from their choices to refine the algorithm and show more accurate recommendations. The system takes into account various factors such as price, location, and amenities, as well as user preferences for size, style, and other features. It aims to make the process of finding a new home more efficient and personalized, ultimately resulting in better matches and happier customers.

How we built it

We first cleaned up and preprocessed the database so it was more usable. Then, we implemented a feature calculation algorithm to refine all of the houses the customer liked. Also, we made use of the Gower algorithm to calculate the distance to all other instances and find the best match. About the GUI, we first used the Qt framework and designed a simple, user-friendly interface that allowed users to quickly provide feedback on their preferences, but further on we decided that a Django approach with a database was more usable.

Challenges we ran into

As everyone knows, hackathons are an amazing thing and are usually stressful because of the lack of time you have in order to finish the project. The main challenge was time, and because of that, our idea could not be developed as we want it. Another is the use of new technologies we decide to use on that project. We all were new to the topics of deep learning such as training models, GANs, and Pytorch. This feature did not work on time so we have not posted it.

Accomplishments that we're proud of

Our best accomplishment, every time we attend a hackathon, is to finish our project in order to show what we had in mind to the judges. We also are so happy to learn new technologies to improve our skills in problem-solving and learn new possible methods and have a little piece of our idea to show to others.

What we learned

We learned a lot about teamwork. We also learned about GANs, Diffusion Models, Conditional GANs, Space Latent, Pytorch, and a lot of Python for full-stack development.

Built With

Share this project:

Updates