Kaiser's rule is the rule frequently used in the principal components analysis for selecting the suitable the number of components. When the components are derived from correlation matrix of the observed variables, the rule advices retaining only those components with eigen values (variances) greater than one.