Forecasting of the Internet Usage in the World Regions

Abstract

The paper focuses fitting of different forecasting methods based on annual time series data for the Percentage of individuals, aged between 16 and 74 years, using the Internet, for eight World regions. For comparison, European Union and the World are used, too. For the 2017 and 2020 forecasts, seven forecasting models were applied: three naïve models, geometric mean, simple average, linear trend and exponential trend model. The minimum mean squared error (MSE) indicated the best fitting forecast model for each of the regions. Forecasting models based on the historic data period from 1996 to 2015 indicate the new ranking of the World regions regarding the Percentage of individuals using the Internet in the near future, e.g., an increase of Percentage of individuals using the Internet in Arab World from 0.04% in 1996 to 40% in 2015, with the forecast of 60% in 2020, fixes this region at the last rank among considered eight regions, but North America, with an increase from 15% in 1996 to 76% in 2015, with the best fitted forecast of 83% in 2020, moves from the first rank in 1996 to the third rank in 2020, after Europe & Central Asia and Latin America & Caribbean. If the forecasts for the observed indicator would be based on a shorter past time series, from 2006 to 2015, in 2020 a different ranking of the World regions would appear.