ESTIMATION OF A REGRESSION POLYNOMIAL

A classification of the statements of regression problems is discussed . The problem of finding the information subset of predictors is one o f the most important. A model problem for this problem is that of esti mating the power of the polynomial in a pairwise regression. The resul ts of an asymptotic investigation of a series of estimates of the poly nomial power are presented. Problems of comparison of the statistical methods are considered, and the application of the methods of statisti cal tests, in particular. of the bootstrap, is demonstrated.