Methods for daily forecasting

Abstract

For many branches of business daily forecasting constitutes an important activity, since future plans are made or decisions are taken according to the future expectations. This thesis provides a comparison of different daily forecasting methods with special implementations for different time series data and evaluates the results of forecasting experiments. It is important for the selected forecasting method to give more accurate results and smaller forecast errors. Also, the coding and the implementation processes have an effect in choosing a forecasting method. The first chapter explains forecasting, the second chapter gives information about the forecasting process, the third chapter gives examples of the forecasting methods used in the literature. The explanation of forecasting methods begins with the fourth chapter with the explanation of the so called "naive method". The fifth chapter explains time series regression methods, the sixth chapter presents exponential smoothing methods, the seventh chapter gives information about the autoregressive integrated moving average method. The eight chapter explains the implementation of these forecasting methods written on "R" program. The experiments evaluating these forecasting methods are introduced and explained in the ninth chapter. Finally, a general overview of the methods is presented.