Colibros

Inspiration

We picked this challenge, as gathering live data from a train on site seemed like the most hands-on experience.

What it does

Creates monthly summaries of data collected from trains on the DB network, giving the means to compare and find relations.

How we built it

We combed the endpoints to understand the structure of data, and gather information on how to draw the best conclusions from the limited amount of information available. After designing an algorithm, we used python to manipulate the data.

Challenges we ran into

The API's structure was challenging to navigate, along with lack of documentation, bugs, and lack of descriptive error messages. During the night the server shut down, causing delay in the development.

Although the DB staff was really friendly and approachable, they couldn't provide us with enough technical information. The challenge description wasn't clear enough to us, and changed as API shortcomings came to light.

The amount of data we were provided wasn't sufficient to train a model, therefore we could only end up with a simple data extraction tool.

Accomplishments that we are proud of

Despite the challenging environment, we managed to deliver a product that can be used as part of a more comprehensive solution.

What we learned

We understood how bad API design can slow down the development process, and how important clear documentation is to make communication between the parties involved more efficient.

We learned how to adapt to a changing environment, and make the most of limited resources, even if the goal is unattainable.

What's next for Colibros

We would like to provide an outside perspective of the Colibri API, sharing suggestions, making light of shortcomings, and providing additional testing.