Evaluation - simula/biomedia-2020 GitHub Wiki

For the evaluation, we will use mean squared error, mean absolute error and the mean absolute percentage error for first two challenges. For the ranking mean absolute percentage error will be used which shows the improvement of the automatic prediction compared to the average prediction. For the optional challenges, we will use manual evaluation with the help from three different experts within human reproduction.