Which statement accurately describes the impact of target leakage during the preprocessing phase?

Question

Data Science — Hard

Which statement accurately describes the impact of target leakage during the preprocessing phase?

Accepted Answer

Target leakage occurs when information from the target variable is inadvertently included in the features used for training. This allows the model to 'cheat' by using the target to predict itself, resulting in artificially high training performance. However, this relationship does not exist in real-world production data, leading to a massive performance drop when the model is deployed.