Data Science Challenges; A Whole Systems Lens for Energy Network Solutions

The rapid digitalisation of the energy system requires advanced and robust data driven solutions as well as the necessary technical skills to implement them. Unfortunately, many solutions are only tested on private data using a single method only compared (if at all) to simple, non-competitive benchmarks. This approach limits the reproducibility and usefulness of the outputs. This report demonstrates, through three distribution network focused problems, the opportunities, and limitations of data science competitions in supporting open science innovation. With each challenge this report includes a description of the problem, the data utilised, and the solutions given. Through the outputs, it is demonstrated how different models compare (including against competitive benchmarks), how to best utilise the available data, and what unique techniques can help to optimize the solutions. The paper concludes with some of the limitations and possible improvements in future competition implementations.