NOTE: the first few prediction models obtained in the earlier iterations (known as the \(burn-in\) period; denoted by \(L\)) are typically thrown away since they tend to not provide very good results, like you throw away the first pancake of the batch
