On the Selection of Growth-Curve Models for Forecasting the Diffusion of IT Technologies

Abstract

Several growth-curve forecasting models are compared with respect to empirical data sets from IT industry of Korea and attempt is made to suggest a case-wise guiding principle in terms of technology (service) type and data length. It turns out, as expected, no single model consistently outperforms others across all data sets. Instead, for a given data type, we can

determine which models should be recommended and which should be avoided. A good practice then is to evaluate two or three preferred models in terms of multiple performance measures and make a final decision based on the results. The practical usefulness of the select ion guideline is confirmed by practitioners of some principal IT firms in Korea.