Evaluation of the Quality of Water Using Fuzzy Reasoning and Multiple Regression Analysis

Abstract

We investigated the suitability of the palatable water standards established by the Ministry of Welfare in Japan using chemical analyses and sensory tests data on 82 water samples obtained in Hiroshima and other districts. We confirmed hjgh correlations between the results of sensory tests and four explanatory variables (organic matter, total handness, iron, bicarbonate), indicating appropriateness of the standards. Furthermore, we constructed multiple linear regressions and simplified fuzzy models using the four factors as explanatory variables. It shows that simplified fuzzy models are more practical than multiple linear regressions.