The evaluation metric for this competition is RMSE . The Root Mean Squared Error (RMSE) tells you something about the accuracy of your predictions. The higher the RMSE, the further your prediction differs from the truth For every athlete in the dataset, submission files should contain two columns: id and predicted. id should be set as the index column of your submission file. and predicted contains the predicted finish times for all athletes in the test set. Make sure your submission has this format, otherwise you will not be able to upload your submission to Kaggle.