Comparison of Demand Forecasting Methods for Global and Local Demand: The Case of Classic Literature Demand Forecasting

Abstract

Forecasting is an integral part of entrepreneurship. The aim of this paper is to use Google Trends data to predict the local demand for books in the Czech Republic and compare it with the global demand. The forecasting will be done with data by Google Trends and by book purchases. Both approaches will be compared. It will also be evaluated whether the use of multiple measures of accuracy will lead to different results. The methods chosen for forecasting were seasonal naive method, ARIMA and ARFIMA method, ETS method, Holt and HoltWinters method. The calculations will be completed with BATS and artificial neural network methods and Hybrid method. For the GT data, the extent to which the search-based model accurately matches the actual purchases was quantified. It is an innovative concept. From the results, it can be concluded that all methods on the data for the Czech Republic, and on both sets of data, predict demand stagnation. For the global comparison, the results are different. Furthermore, it is clear that the calculation by search I purchases data gives similar results.