The application of factorial analysis in the classification of the attributes of real estates

Abstract

The base for differentiating factors comprises proper values determined based on the correlation matrix between the variables, i.e. variances differentiated by the subsequent fac­ tors. Based on this measure it is possible to establish how many factors are important from the practical point of view. The most commonly used method is Kaiser criterion. This crite­ rion leaves only these factors that have their proper values bigger than 1. This means that if the factor does not differentiate at least as much as one original variable - this factor is re­ jected. The second method applied in the paper is a graphic talus test, where proper values are presented on a linear graph and then the place is defined in such a way that to the right from this place there is a mild fall of proper values. To the right from this point there is probably only a factorial talus. The application of the Kaiser criterion mak.es that, in some cases, there are too many factors left, while the talus test usually leaves too few factors. Both methods give similar re­ sults when there is a high number of observations and small number of factors. In the pa­ per both methods were applied. To verify if the differentiated number of factors was suffi­ cient, the matrix of residual correlations was made. They enable us to find possible lacks in adjustment; it means to observe which correlation coefficients cannot be properly reconstructed by a given number of factors.