You can open the demo here
It all started simple enough: Can we predict how much is a property going to cost before it is even built?
Surprisingly the answer was “yes”.
We got our hands dirty into as many data sets as possible, everything from census, demographic and population data to financial indicators, school performances and even how many McDonalds are in the neighborhood. Armed with so much data and old fashioned machine learning algorithms (can’t tell you which ones, confidential stuff) we managed to generate sale prices for each neighborhood with a granularity down to how many rooms is the future property going to have.
Generally we can predict prices one year in advance, but the target is to work it to ten.
Anyway, visualization was based on Google Maps, it is a bit heavy on the client side, but works well a s a simple demo.