Frustration at train lateness and long delay times.

What it does

Analyzes Amtrak's database of train arrivals, departures, and delays in order to predict a train's chance of being late to help decide if you should take an alternate form of transportation.

How we built it

Amtrak Status Maps Archive Database has an in-depth archive of trains and when they're late. By lifting the data from a period of time, formatting it, and running it through a feed-forward Neural Net, we could predict if a particular train was likely to be late. We then created art assets and a faux phone app UI to represent how the finished product would look.

Challenges we ran into

Amtrak's archive was difficult to take data from. Their data formatting was tedious to work with, and their methods of data organization changed over time. They also block IPs that appear to be taking data, so future implementations will have to work around that issue.

Accomplishments that we're proud of

Successful data formatting regardless of the difficult and changing data organization methods.

What we learned

We all had experiences with new software and concepts, such as TkInter, TensorFlow, and UI design.

What's next for Amtracker

Future app concepts include:

  • A list of future stops for the particular train and the projected delay
  • A "Plan your Trip" option which gives you other transportation options nearby if your train is delayed
  • A real-time map of the train

Built With

  • amtrak-status-maps-archive-database
  • perl
  • python
  • tensorflow
  • tkinter
Share this project: