Multi-chart schemes for detecting a range of variance changes

This article considers how CUmulative SUM (CUSUM) and Exponentially Weighted Moving Average (EWMA) multi-charts were used to monitor the variance (variability). Multi-chart schemes are a combination of several single charts that detect changes in a process quickly. We give a theoretical equation to arrive at the optimality of the CUSUM multi-chart for the variance. Simulation studies and asymptotic analyses show that the CUSUM multi-chart has better performance than the EWMA multi-chart in detecting variance shifts in an i.i.d. normal observations. The variance change problem has diverse and attractive application areas; we give a real example in the field of agriculture that monitors the price variability of maize to demonstrate the practicality of the schemes.