Time Series Forecasting for Clients Rates in Tele-Communication Data using Statistical Techniques

Abstract

Nowadays, massive data highlights the significance of exploiting data mining technology for business needs. Data estimation is one of the business important demands, which is one of data mining objectives as well. Hence, data mining has utilized machine learning (ML) and statistical analysis (SA) techniques for developing business intelligence solutions. Time Series (TS) forecasting methods are tested and enhanced. Such enhancement is increasing the power of TS abilities, which is served to respond to business future estimation requirements. In this paper, an investigation for TS methods distinction is implemented. Four TS methods are applied to forecast the next five years of clients’ rates of two Yemeni’s communication companies. The four methods are Movement Average (MA), Weighted Movement Average (WMA), Least Square (LS), and Exponential Smoothing (ES). The estimated data is evaluated by comparing TS series forecasting methods.