We wanted to see if we could identify travelling trends through twitter, and use that in a meaningful way. This led us to the idea of figuring out whether travellers tend to go to 'mainstream' destinations, or more 'hipster' locations. We decided to use this to make a travel destination recommender, in collaboration with a lot of other data.
What it does
Examines the user's travelling trends (where they travel, what time of the year they travel, how much they spend on their trips, etc) in order to make suggestions for their next holiday.
How we built it
- To identify popular travel destinations at any given time in the year, we built an event-detection system that analyzes tweets by classifying and clustering tweet-features, to determine periodic events.
- To get information about the users we used the Deutsche Bank API
- To get flight information we used the SkyScanner API
- To retrieve cool pictures of every destination we used the Bing-Image-Search API
- Used the bottle framework to build the system
- Angular, bootstrap for front-end
Challenges we ran into
- Very technically difficult problem to use machine learning algorithms to identify travel events
- Had to come up with clever tricks to handle rate-limiters on APIs
What's next for DBTravel
- Get train information by using the RomeToRio API