Spatial and temporal multivariate statistical analysis to assess drinking water quality in medical services

Abstract

Drinking water quality supplied to medical services presents significant role regarding the health aspect of the society. Multivariate statistical techniques were applied for the interpretation of data obtained, i.e., cluster analysis (CA), principal component analysis (PCA), factor analysis (FA), and discriminant analysis (DA) to analyze and assess the spatial and temporal variations of drinking water quality in different medical services in Kuwait. This study was generated over a period of 11 years (2007–2017), including 19 parameters at fourteen different sites. Hierarchical CA obtained two groups regarding both spatial and temporal variations. For spatial variations, 14 sampling sites were grouped into Low Concentration (LC) and High Concentration (HC). For temporal variations, 12 months were grouped into Summer and Winter. DA provided better results by data reduction for the large data set with great discriminatory ability for both spatial and temporal variations, as only five parameters were used concerning the spatial variations to afford 68.4% of the cases being assigned correctly, and seven parameters were interpreted for the temporal variations affording 76.1% of correctly classified cases. The applied PCA/FA on the spatial variations resulted in five principle components (PCs) for the LC region, and the total variance is 74.84% and three PCs for the HC region explaining a total variance of 64.86%. For the temporal variations, summer yielded into five PCs with a total variance of 70.6%, whereas the winter resulted in three PCs describing 67.1% total variance. Thus, multivariate analysis provides better spatial and temporal variations assessment in contemplation of effective drinking water quality management and control.