Application of Exponential Smoothing Models and Arima Models in Time Series Analysis from Telco Area

The use of ICT has been steadily increasing in both business and social life. Apparently, the most essential communication technology today is a mobile phone. However, the use of mobile phones also has its downside, which is their impact on the environment, which is not negligible. Despite the negative impact on the environment, few of us can imagine a life without a mobile phone. The mobile telecommunications market is one of the most important sectors of the modern economy. Analysing past developments - as well as predicting future developments - of indicators in this area plays a very important role in decision making, as in any other area of the national economy. The extrapolation methods have been the most often applied methods in the area of time series analysis and forecasting in practice. Currently, the combined methods in time series forecasting is more and more favoured. The main aim of this paper is an examination of applicability of the Box-Jenkins methodology models and the exponential smoothing models for providing extrapolation forecasts, but for past development modelling of selected indicators from the telecom area, too. Information on the indicators under study was collected on monthly and daily basis. Quality of the models selected was then assessed using the MAPE and AIC metric. In conclusion, a comparative analysis was performed of both the groups of models. The best individual models were further aggregated and quality of these was assessed using the same assessment criteria. SAS statistical system was applied for effective implementation of the analysis. The research has demonstrated that the exponential smoothing models can only be recommended for the analysis of indicators under study from the mobile telecom area. The detailed analysis has proved, anyway, a higher level of success with the combined models.