ANALYZING THE SUITABILITY OF TIME SERIES AND REGRESSION FORECASTING METHOD FOR DRINKING WATER PRODUCT

Abstract

The objective of this research was to find the right method to increase the accuracy of
forecasting in the demand of 330 ml bottled drinking water in the water supply company in
Bandung West Java, Indonesia. The methods used were Naïve, Moving Average, Weighted
Moving Average, Single Exponential Smoothing, Holt’s Exponential Smoothing, Winter
Holt’s Exponential Smoothing, and Polynomial Regression. To examine the performance of
forecasting methods, three measures are applied. Those are Mean Absolute Deviation
(MAD), mean squared error (MSE), and mean absolute percent error (MAPE). The research
results showed that the most suitable forecasting method for 330ml bottled water product is
weighted moving average. By comparison MSE resulted from proposed method is
1,334,992.74 and the number of MSE with the method that is currently employed by the
company is 2,382,366.84. The conclusion is the accuracy calculation of the demand for
bottled drinking water products can be done by techniques forecasting with weighted moving
average method.