Horrible UX in current apps for apartment search. Lack of proper personalisation and lack of unification in the european real estate classifieds websites.
What it does
It aggregates classifieds supplied by different real estate agencies and shows relevant classifieds to users, according to their personalised preferences.
How I built it
We used PHP (+ Laravel) for the backend of the application and MySQL for the database. Both are hosted on the Amazon EC2. For the machine learning bit, we wrote a Python script, which runs a basic collaborative filtering method to show relevant classifieds. The front end is an iOS app written in Swift. We also used the Apple maps API and the Google places API
Challenges I ran into
The collaborative filtering only works properly if there is enough relevant data stored in the database already, because it operates by comparing and finding users closest to each other and suggesting classifieds based on that criteria. Since we had a very short timespan, we only had a handful of time to add some users and classifieds.
Accomplishments that I'm proud of
The integration between three loosely coupled parts of the application. The PHP backend is connected to the Python machine learning part and the Swift front end part.
What I learned
Hackathons are awesome. Integrating different technologies is quite difficult. Sleep is important. Sleep...
What's next for SMARTY APARTY
Research the market and competition and challenge ourselves with a few startup events.