各國電信開放以來,各國電信產業進入快速變遷的時代。通訊系統由過去大家所使用的2G系統逐步邁向新一代的3G系統;但是系統業者在建構3G系統之際,必須仔細考量消費者的轉換意願以及採用程度,再者3G系統的建置資金十分龐大,因此對於系統業者來說,如何建構一套模型來預測採用者人數變的相對十分重要。 本研究採用多代擴散模型以及時間序列方法來做預測工作,研究對象為日本2G及3G的採用者人數加以探討。資料來源為日本電信協會行動電話資料庫,採用時間點為1997年4月到2006年5月;並加以對日本三大電信公司分別作探討以及總採用人口做分析,並驗證其分析能力。分析工具為SAS,並採用非線性迴歸最小平方法做估計,並以SPSS作時間序列分析。 本研究透過兩個不同的多代擴散模型針對電信業者做分析後,發覺Bass在2004年所提出的多代擴散模型有較高的配適度;而透過時間序列分析則發現時間序列分析法配適度相當的準確,但是若預測期數略為增加,則整體預測程度下降相當明顯。
As the opening of the communication industries, the environment had a rapid change. The 2G system are continually evolving into 3G. When building the 3G system, the vendors had to considerate about the potential that the consumers would take because of the huge fund. As to the system vendors, how to build a forecasting model is very important. The research used multi-generation diffusion model and time series to forecast the subscribers. The data is taken from Telecommunications Carriers Association subscriber database in 2G PDC and 3G WCDMA subscribers during the period of 04/1997 to 05/2006. The analysis included three Japan communication company, NTTDoCoMo, KDDI, Vodafone and the total subscribers of the Japan Communication market. The analysis tool were SAS and SPSS. The result indicate multi-generation with different model, the BASS(2004) model had the highest accuracy. And the Time Series had indicated that if the period of the forecasting is longer, the accuracy will fall obviously.