A machine learning model's objective is to model our dataset. In order to assess each model's performance, we define an objective function. These functions usually express a cost, or how far from perfect a model is. These cost functions usually utilize a loss function to assess how well the model performed on each individual dataset instance.

Some of the most widely used cost functions are described in the following sections, assuming that the dataset has n instances, the target's true value for instance i is t_i and the model's output is y_{i .}