[KongchangAI]
Concept数据泄漏 / 信息泄漏

Feature Leakage

The phenomenon in machine learning backtesting in which improper use of future information by the feature pipeline causes inflated model performance, commonly seen in time-series forecasting when the temporal availability of features is mismanaged.

Timeline (last 90 days)

Nov 10

If actual observed values (such as actual temperatures) are mistakenly labeled as known future covariates, the model will "see" future information, causing backtest performance to be inflated.

Unverified50%

All Facts (1)

Source Articles