| Summary: | 碩士 === 國立東華大學 === 企業管理學系 === 102 === The purpose of this research is to find an efficient method of Time Series Analysis forecasting for bottled water consumption in top twelve countries. Adopting a bottom up approach, it attempts to identify the most suitable forecasting method for each of top twelve countries, separately. Therefore, the individual countries’ bottled water consumption figures for the period (1999-2011), form the basis of the analysis. Six different Moving Averages (MA) methods along with the Exponential Smoothing and Linear Regression methods are applied to historical data in order to forecast the market. Two kinds of measurement errors as RMSE and MAPE are used as a basis of evaluating and comparing the accuracy of various forecasts and then the most efficient forecasting method is identified. Finally, it is noted that, in view of the numeracy of factor affecting the market, the best forecasting method for each country does not necessarily remain the same from one year to the next. Therefore, the objective of this study is to provide better insights into the practice of forecasting in an industry as encouraged.
|
|---|
