The Time Series Analysis Study of Broadband Development-Forecasting Model

• Main Authors: Ping-Huang Tung, 童炳煌
• Other Authors: 謝俊宏
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/a7g528

碩士 === 國立臺中科技大學 === 企業管理系事業經營碩士班 === 103 === As the technology we use daily becomes more advanced, the broadband has become the important tool obviously. Of course, the broadband growth and development will bring the great benefit to a country in the future. The data of the study is from National Communications Commission(NCC). First, collect the number of broadband account over the past decade. And then, divide the “Broadband” into “Broadband” and “Mobile Broadband” to do the forecast research. Finally, is based on the principle of “The Time Series Analysis Study” to build the forecast model. The research method is to use ADF test, JB test, Q test and Q2 test to diagnose the forecast model. And then, checking the forecast model belongs to ARIMA, ARCH or GARCH. Finally, is based on the MAPE principle to measure the performance of forecast model. The research results are summarized as below: 1. The broadband and mobile broadband exist non-stationary, so all of the data need to do the “first difference” before “The Time Series Analysis” process. 2. After forecast model analysis, the Broadband forecast model is GARCH(1,1), and the Mobile Broadband forecast model is ARCH(1). 3. The research result of the MAPE(Mean Absolute Percentage Error) value of Broadband and Mobile Broadband are “acceptable model”. Suggestion: The Broadband and Mobile Broadband are acceptable forecasting model. They could be used for forecasting in the future, but the original data is limited. For further research, still need to keep collecting data to do the better forecasting model.