4.2. Factor analysis
Factor analysis is a multivariate analytical technique, which derives a subset of uncorrelated variables called factors that explain the variance observed in the original dataset (Anazawa and Ohmori, 2005; Brown, 1998). Factor analysis is used to uncover the latent structure of a set of variables. In technical terms, common factor analysis represents the common variance of variables, excluding unique variance, and is thus a correlation-focused approach seeking to reproduce the intercorrelation among the variables. On the other hand, components (from PCA) reflect both common and unique variance of the variables and may be seen as a variance-focused approach that reproduces both the total variable variance with all components as well as the correlations. PCA is far more commonly used than principal factor analysis (PFA). However, it is common to use “factors” interchangeably with “components” in multivariate analysis.
Factor analysis can be performed on any kind of scientific data to establish a pattern of variation among variables or reduce large data sets into factors for easy handling and interpretation. The total number of factors generated from a typical factor analysis indicates the total number of possible sources of variation in the data. Factors are ranked in order of merit. The first factor or component has the highest eigenvector sum and represents the most important source of variation in the data. The last factor is the least important process contributing to the chemical variation. Factor loadings on the factor loadings tables are interpreted as correlation coefficients between the variables and the factors.
In this research, PCA was applied to chemical data from the Ankobra Basin to extract the principal factors corresponding to the different sources of variation in the data. Here, PCA was selected for the reasons stated above. In order to maximize the variation among the variables under each factor, the factor axes were subsequently varimax rotated.