The evaluation metric for this competition is
[Root Mean Square Error.](https://en.wikipedia.org/wiki/Root-mean-square_deviation)
RMSE is the most widely used metric for regression tasks and is the square root of the averaged squared difference between the target value and the value predicted by the model. It is preferred more in some cases because the errors are first squared before averaging which poses a high penalty on large errors. This implies that RMSE is useful when large errors are undesired.