This is the submission of Group 97. Alisha Kiefer (s241870), Florian Dirnberger(s242022), Jason Quinn(s232735), Raimo Sieber(s242186)

Since different approaches were developed we sticked to different Branches to work independently. Here is a quick overview of the important branches:

• Main: First parametrized Model and Wandb setup and Baseline performance
• FloFlo: Time series neural networks
• Jason: Data Preprocssing Approaches and Optimized CNN
• attention_layer: Optimized CNN and Sweeps in Wandb

In the main branch there is the requested Jupyter notebook. Please consider that due to the NDA we signed with Trackman and Mark, we only used randomized dummy values to show that the models work. Test errors can significantly differ from our actual results.