Demand forecasting of high-speed Internet access service considering unknown time-varying covariates

In order to forecast the demand for information and communication services, it is important to consider not only intrinsic variables representing service characteristics but also unknown time-varying variables such as marketing policy. However, in many cases, information such as a company's internal marketing policy is not available. This study proposes a negative exponential growth curve model that incorporates unobservable time-varying covariates by reversely estimating the unknown covariates. The proposed approach is then applied to technological forecasting of high-speed Internet access services provided by a telecommunication corporation in Korea.