| Abstract: | Change Analysis "in the strict sense" is concerned with the problem of detecting and estimating slow and abrupt changes in the probability distributions of successive observations Y(t) of a variable or system. This paper has two goals (1) introduce an approach to Change problems by introducing analysis of Score Change Processes (whose idea is to study if a model to a whole data set fails to fit it by "random walking" the parameter estimating equations); (2) develop analogies between four basic statistics problems, corresponding to the standard assumptions made about a sequence of observations Y(t), t = 1,...,n; test the hypothesis: A: Distribution of specified parametric form, B: Independence, C: Identical distribution, For a sequence of bivariate observations X((t), Y(t)) one would like to test D: Independence of X and Y. Contents are: Introduction, Change analysis in the strict sense (test Assumption C), Goodness of fit (test Assumption A), Spectral Analysis (test Assumption B), Four phases of change analysis, Parametric scores change analysis, Nonparametric scores change analysis. |
|---|