EWMA/CUSUM

EWMA and CUSUM are time-weighted control charts designed to detect small, persistent process shifts that traditional Shewhart charts often miss.

While standard Shewhart charts rely on single data points to flag major spikes, EWMA (Exponentially Weighted Moving Average) and CUSUM (Cumulative Sum) charts utilize historical data to identify subtle drifts. CUSUM works by accumulating deviations from a target value (summing the 'errors' over time) to signal when a process has moved as little as 0.5 to 1.0 sigma. EWMA applies an exponential decay factor (typically a lambda of 0.2) to give recent observations more weight while still accounting for the past. These tools are the industry standard for high-precision manufacturing and automated process control: CUSUM excels at pinpointing exactly when a shift occurred, while EWMA provides a smoother visual representation of the current process mean.

https://www.jmp.com/en_us/statistics-knowledge-portal/exploratory-data-analysis/cusum-and-ewma-control-charts.html