If your complexity on the product is greater in reaction, then the training error decreases. But if the hypothesis is simply too intricate, then the product is matter to overfitting and generalization will likely be poorer.[43] Their level of popularity has reshaped the tech market