A qualitative approach to the change detection problem with different types of sampling

In this paper, we are interested in nonstationary processes. We propose a method for change detection, based on a test of residual balance. We present some algorithms allowing one to take into account the different types of sampling procedures (simple, multiple, sequential, slippery) whose advantages and limits are underlined. The choice of the sampling type and of the size of samples is based on the calculation of wrong decision risks.