chapter  12
58 Pages

Statistical Change-Point Analysis

ByAlbert Vexler, Alan D. Hutson, Xiwei Chen

In health studies there may be a distributional shift, oftentimes in terms of a shift in location either due to random factors or due to some known factors at a fixed point in time that is either known or needs to be estimated. This chapter outlines the general aspects of change-point problems and the variety of methods to deal with change-point data analysis issues. It presents parametric, nonparametric, and semiparametric change-point detection methods in a variety of model settings, including the single change-point problem, the epidemic alternative, which is a special type of a multiple change-point problem, and change-point problems in regression models. Regression analysis is applied in many disciplines, including clinical trials and medical studies. In practice, ignoring changes in the data structure may lead to a poorly fitting regression model, which in turn may lead to incorrect inference. The statistical literature has several research studies of the problems associated with regression models.