The goal of this project will be two step; the
ultimate goal is to increase the prediction accuracy. Given the amount of data available this will normally be challenging. The second goal will be to develop and test methods for enabling the first goal: The students will be testing new methods that helps state of the art ML methods perform better on small data-sets. Ideas from
applicants are welcome but some starting points could be:
- Discretization (moving from regression to classification as target variable is
originally a continous variable).
- Data expansion via GAN.