While home for the summer Lucas went to the park with his friends to play frisbee. Going with a group of 7 friends, they luckily ran into 7 other people at the park looking to play. They continued playing every Monday night and the group expanded to around 40 students around their community. These new friendships would have never been possible without meeting up to play.

The thought is if there is an application that raises the odds of people meeting up, then there are higher odds for people to make new friends in their community.

What it does

GitActive uses Canada's Open Data API to get information regarding parks close to your area. Users would create events for pickup games, and people would join in. We also implemented the Yelp API to highlight restaurants close by for post sports celebrations. Machine learning was also applied to predict and create future events faster and better.

How We built it

We built our hack using PHP, JavaScript, HTML5 and CSS hosted on AWS with leaflet.js as my go to the map. We used PHP to handle the API and to perform POST and GET requests. For the Machine Learning component, we used AzureML to perform our prediction; where we trained the machine unsupervised.

Challenges We ran into

One challenge we ran into was how to build the machine to predict the dates. We originally thought Amazon Machine Learning would be the way to go along with AWS as our cloud. Unfortunately, AML was not available on the free tier. We went straight to Python and Flask next, but all of us had no clue how we can train the machine effectively to target a few of our dataset. Next, we looked at Microsoft Azure Machine Learning, we were able to successfully train our machine. Even though we were able to train the machine, we had trouble at times with Azure. For one, we were trying to test out the API call using Python. According to a few documentations, we should use the API key, which makes sense, but when we ran it on Python 3, we got 401 error. Unfortunately, the Microsoft people were gone for the night, so we could not ask for their awesome help. After 2 hours of trying, we found out that we should have used the primary key provided instead of the API key provided, and we were able to perform a POST request on Python. Another problem was trying to perform an HTTP request with AzureML to PHP. Unfortunately, we must use an HTTPS request in order for us to be able to apply machine learning to our hack. We were also short on time, so we decided to leave our trained machine behind.

Another roadblock we encountered was using git. As Nicole and Brooke have not used Git before, files were overwritten accidentally, pushing us backwards. This was also their first hackathon and we focused on learning new things instead of getting out the largest application.

Accomplishments that we're proud of

  • With none of us having any experience working with Machine Learning before, we were able to train a machine to successfully predict stuff for us
  • Nicole and Brooke survived and had a blast at their very first hackathon
  • A successful product!!!!

What We learned

We learned a lot this weekend. We all learned a bit on Machine Learning especially Raymond which was his main priority for the hack. For Brooke and Nicole, they had a blast learning about web development, PHP and using APIs to make a request from Lucas's API. Lucas also learnt a lot about web hosting, spinning up his first AWS node.

What's next for GitActive

Our goal is to improve this platform by implementing the finished ML before launching. We also hope to use user location to further advertise nearby restaurants.

Share this project: