Serious game & paper: Hydro Hero

With a group of five, I developed a so-called serious game for children to learn about water management, especially water quality. In the game made for a course, called Hydro Hero, the protagonist has to remove all the disadvantageous matter from the water whilst leaving in the benificial of neutral matter. While figuring out what matter belongs in which catogery, children learn about their effects on the water. The reason behind this, is also taught by letting breaking up the game with minigames.

You can play the game on the website of a group member of mine.

We tested the game scientifically and proved that children indeed learn what belongs in the water and what doesn't from this game and we also found out that they very often knew the reason why this was the case. After the course, with the help of some others, I wrote a paper about our findings, which got published in conference proceedings of a education game alliance.

MLFlow Collaboration wrapper

For my job, I wrote a wrapper around MLFlow, which is a tool that allows for easily logging machine learning experiments. I found out that there are some drawbacks to MLFlow, of which I tackled the following:

1. 1. Collaboration on a central locations is quite difficult with the original package.
2. 2. Setting up the logging could be made easier and quicker.
3. 3. Basic logging of standard packages could be made even easier.

In the package, called MLFlowCollab, setting up basic logging is way easier and saves a bit of time, collaboration is made way easier if a shared folder is available and scikit-learn logging has been turned into a oneliner, while keeping a lot of flexibility.