In statistical modeling, the mistake of fitting a model too closely to the training data, leading it to perform poorly on unseen data.