Regularization comes into the picture when a model is either overfit or underfit. It is basically used to minimize the error in a dataset. A new piece of information is fit into the dataset to avoid fitting issues.
