Discriminant Analysis of Water Quality Data in Langat River

Abstract

Highlights ► Cluster analysis (CA) using single linkage method, SLM on the un-centred correlation distance was considered to group the most similar internal structure between objects. ► Thirteen variables were found to be significant variables based on stepwise discriminant analysis, SDA. ► Canonical discriminant analysis, CDA revealed that13 variables were sta-tistically significant in discriminating among the sites cluster. . Abstract River monitoring processes usually generate large databases of water quality variables. The water quality variables generated from the monitoring pro-cess are important to identify the status of the river. Due to multidimensionality of complex characteristics in river water, meaningful information from a large data-base can be extracted by using multivariate statistical methods, i.e. discriminant analysis, DA. Appropriate univariate transformation and data screening for multi-variate outlier's detection were considered. The DA approaches used in this study gave better information on river water quality, especially concerning the contribu-tion of the variables in discriminating between the three spatial areas in Langat River.