The Clarifai API and a lot of thinking by two of our members. We wanted to work on developing a AI that could analyse a problem way better than a human. The real estate industry was for us an awesome gateway to this challenge because it's a market where the expertise is owned by few peoples and the general public is easily afraid of it.

What it does

It analyse data provident on a real estate property and compare it to all the data we got on passed transactions. It can then estimate the price of this property within a certain margin. This way, the general public and real estate agents could use this tool to have an idea of the net worth of their property or the one they prize.


How we built it

We user Azure's Machine Leaning lab to train an AI(model) to be able to estimate real estate value. It can be accessed by our webservice from :

We also used AWS as it can provide a way-point to all the information we gathered on real estate sites and can provide our web application.

Tech Talk : (AWS Running Windows Server on EC2 with a RDS Database linked with the Entity Framework ORM. We also have Beenstalk deployment integrated to our IDE (visual studio) and a Route 53 reroute from our public DNS to the provided .tech domain. We also had a S3 bucket to store images files to use with the Clarifai APi)

We also use the Clarifai API to get tag from images and compare those values with the ones entered by users. We used a lot of models and training to find the right one who offer the lower error on our data.

Challenges we ran into

Obtaining the data was one of the challenge. Learning to use the Machine Learning lab was also a big one. We were able to make those two thing done, even if they took more time than planned, and get persistent and consistent results.

Accomplishments that we're proud of

We were able to build a successful Machine Learning Api that is able to crunch data as fast as it can be provided.

What's next for Real Estate Genius

Getting more data is our top priority. With more data, we can refine our AI and offer even more precise result. Including other area than Montreal is also high on our list.

There are so many open API that provide information on real estate property and their surrounding. Montreal offer data that could be useful to improve our AI. Other information than real estate price could also be crunched. Hydro-Quebec bill and Condo monthly fees could also be nice information to gather.

Share this project: