Forecasting the Amount of Effluent for a Large Municipal Water Supplier
The Northeast Decision Sciences Institute Annual Meeting Proceedings
The authors obtained effluent (clean water output) data from the Providence Water Supply Board (PWSB) as well as meteorological data for an 11 year period. It was our goal to create a forecasting model for the effluent or discharge for future periods. In addition, we discovered a strong relationship between maximum daily temperature and the amount of effluent, as might be expected. However, we also noted that the relationship was somewhat curvilinear. We further examined the data with a time series graph, an autocorrelation plot, and a partial autocorrelation plot. Ultimately, we wanted to compare linear regression and Autoregressive Integrated Moving Average (ARIMA) modeling as predictive models for effluent data. We also wanted to identify the best predictive periodicity that would fit the water data.