This repo contains code written by Anders Pedersen and Håkon Holte for the Krafthack Hackathon of 2022. The code in its entirety can be found in the notebook named krafthack.ipynb. In order to run the cells in the notebook, the krafthack dataset2 has to be loaded, e.g. from Google Drive or simply uploaded to Google Colab from a local drive. The path to the dataset has to be specified in the cell where this is indicated.
The specifications of the models used to obtain the predictions that were submitted at the end of the hackathon can be found in the "Models" folder. The names of the model folders indicate which of the bolts each model have been trained to make predictions for. The only exception to this rule is the folder for the model used to make predictions for bolt 1, which, due to a small oversight, was named Dense_v4. The name "Dense" refers to the models being fully connected, or dense, neural networks.