Why is cross$-$validation used?

To reduce the size of the dataset.

To assess the model's performance more reliably.

To increase the speed of the training process.

To select the most important features.