因子分析与对协方差矩阵结构的推断 - 副本.ppt

Factor analysis and inference for structured covariance matrices 9.1 Introduction History: Early -century attempt to define and measure intelligence Developed primarily by scientists interested in psychometrics Advent of computers generated a renewed interest Each application must be examined on its own merits The essential purpose: Describe the covariance among many variables in terms of a few underlying ,but unobservable, random factor. Suppose variables can be grouped by their correlations. That is .suppose all variables within a particular group are highly correlated among themselves, but have relatively small correlations with variables in a different group. The primary question: The primary question in factor analysis is whether the data are consistent with a prescribed structure. * * * The factor analysis model: Matrix notation: 9.2 The orthogonal factor model The orthogonal factor model Covariance structure for the orthogonal factor model This ambiguity provides the rationale for “fector rotation”, since orthogonal matrices correspond to rotations of the coordinate system for X. Factor loadings L are determined only up to an orthogonal matrix T. Thus, the loadings both give the same representation. The communalities ,give by the diagonal elements of are also unaffected by the choice of T. 9.3 Methods of Estimation The same covariance matrix S is an estimator of the unknown population covariance matrix If the off-diagonal elements of S are small or those of the sample correlation matrix R essentially zero, the variable are not related, and a factor analysis will not prove useful. If appears to deviate significantly from a diagonal matrix, then a factor model can be entertained ,whereas the major aim of factor analysis is to determine a few important common factor. We shall consider two of the most popular methods of parameter estimation, the principal component (and the related principa