" Data lag " occurs when the model relies too heavily on old patterns that no longer reflect the current reality (e.g., pre-pandemic shopping habits). To mitigate this, " feature weight updating " is used to give more importance to the most recent data points while " decaying " the influence of older data. This allows the model to adapt more quickly to real-world shifts without requiring a full rebuild from scratch. While data diversity and volume are important for general performance, they do not specifically address the temporal " lag " issue like weight adjustment does.