You can choose a challenge for IDEATHON from the provided file of potential topics. As well you can propose your own challenge, but it should be related to AI and contain some new ideas or application novelties. The topic should be aligned with the Desktable.
For the HACKATHON challenge, use these files as a dataset (with data for training and testing). The challenge is to process the data into an ML-suitable format and train the selected model to predict the student’s final grade (round the final grade to 10 joint scale mark) for the course. Then the test data should be used to test the model’s accuracy and performance.
The best solution will be evaluated based on these criteria:
- Prediction accuracy (F1-score, accuracy, recall, precision) – 70%
- Performance (time, needed for prediction, memory usage) – 20%
- Explainability (the ability to provide logic behind the decision-making) – 10%
During the final submission, different data will be provided, but of the same structure as for testing.