以雙階段模糊分段迴歸分析預測多世代行動通訊技術之演變與年專利授權量

Abstract

技術週期 (Technology Life Cycle) 預測在高科技產品的策略規劃和政策訂定扮演重要角色，近年來，雖然多有研究預測技術生命週期，但預測多世代技術生命週期之研究甚少，以專利為基礎，預測多世代技術生命週期之研究更為罕見，由於多世代技術於每一生命週期中的行為類似，專利數量是否也呈現相同模式，值得進一步研究，但相關研究甚少。自1980 年代以來，行動通訊技術進步快速；大約每十年，人們就發展出新一代的技術，使通訊系統不斷完善，並加速全球經濟發展與文明的進步。每一世代行動通訊技術之年專利量變化是否與前世代呈現相同之模式，亦少有學者探討。因此，本研究導入模糊分段迴歸分析，預測多世代技術生命週期已跨越研究缺口。首先，透過前世代 (例如，1G、2G、3G) 行動通訊專利生命週期長度，預測新世代 (例如，4G)技術生命週期長度；其次，透過前世代技術生命週期特定年份之專利數量，發展該年份之模糊分段迴歸方程式，並以之預測未來世代產品於該年度之專利授權量。依據研究結果，預測4G至6G生命周期和5G每年授予的專利數量的準確率分別為74.24％、91.17％、88.57％和91.43％。此外，於多世代技術預測中，模糊分段迴歸分析相較傳統的線性迴歸分析與模糊二次迴歸，有較佳的準確率，發展完善之預測模式，可作為預測各種多世代技術之基礎，也可作為政府訂定創新政策、企業發展策略之依據。Forecasting of technology life cycle is crucial for strategic planning and policy formation of high-tech products. While there has been a considerable amount of research conducted on forecasting the life cycles of products, there is a scarcity of studies focused on predicting the life cycles of multi-generation technologies. Research on estimating the life cycle of multi-generation technology is extremely scarce, primarily relying on patents as a foundation. Given the similarity in the life cycle of each generation, it is interesting investigating if the number of patents follows a similar pattern in multi-generation technology. However, there is a scarcity of relevant research on this topic.Mobile communication technology has experienced tremendous advancements since the 1980s. Approximately every decade, a new generation of technology is established, continuously improving communication networks and driving global economic development and the progress of civilization. There has been limited scholarly investigation into whether the pattern of changes in the number of patents for each successive generation of mobile communication technology is comparable to that of the preceding generation.Therefore, this study aims to introduce fuzzy linear piecewise regression analysis as a method to anticipate the crossing of the research gap in the multi-generation technological life cycle. At first, this research forecast the duration of the life cycle for the upcoming generation (e.g., 4G) based on the length of the technology life cycle for the preceding generations (e.g., 1G, 2G, 3G). Then, this research utilizes the quantity of patents in a particular year of the previous generation's technology life cycle to anticipate the progress made in that year. The fuzzy piecewise regression equation is employed to forecast the quantity of patent authorizations for forthcoming generations of products within a certain year. Based on the analytic results, the accuracy of predicting the 4G to 6G life time and the annual amount of patents granted for the fifth-generation (5G) are 74.24%, 91.17%, 88.57%, and 91.43%, respectively. The research findings indicate that fuzzy piecewise regression analysis outperforms standard linear regression analysis and fuzzy quadratic regression in multi-generation technology prediction in terms of accuracy. The sophisticated and flawless prediction model can be utilized to forecast a wide range of multi-generational technologies. Additionally, it can serve as the foundation for the government to create innovation policies and plans for enterprise development.