Development of multiple linear regression model for biochemical oxygen demand (BOD) removal efficiency of different sewage treatment technologies in Delhi, India

Abstract

Among the various modeling techniques applied to dataset, multiple linear regression (MLR) analysis is the most efficient way to figure out the relationship between the response variable and the predictive variables. This study emphasizes on establishment of multiple linear regression models to analyze Biochemical Oxygen Demand (BOD) removal efficiency for technologies, namely Densadeck, Extended Aeration and Activated Sludge Process. Assumptions of multiple linear regression like linear relationship, multivariate normality, multicollinearity and Homoscedasticity were examined. The data that verify the assumptions were analyzed with multiple linear regression. Time series plots indicate drastic decline in BOD removal efficiency in the month of Feb and March during the years 2012 and 2013. This study was significant as it gives the technology having the best-fit regression equation based upon multiple correlation coefficient (R), coefficient of determination (R2), standard error, residual and F-ratio value. Societal benefits include enhancement in the performance of sewage treatment plants.