科技預測模型之檢測：廣泛的羅吉斯模型、費雪模型、以及甘伯茲模型之比較

Abstract

這篇論文的目的主要是在檢測一個擁有可變動上限以及彈性轉折點的新模型與其他以廣泛應用的科技預測模型之比較。科技預測對於決策者來說是非常好的補助工具，它除了可以用來找尋科技替代的趨勢以及可能的產品發展的上限之外，也可以被用來檢測產品生命週期的可能趨勢。因此，好的預測可以對未來的趨勢提供一個良好的訊息。費雪模型以及甘伯茲模型是兩種廣泛應用的模型，而且現今也有許多改良的模型被應用在各種科技預測的領域。 因此，這篇論文利用預測圖形以及數理指標來檢測這三種模型的適用性。 本篇論文利用了日本家電產品的滲透率來做模型的比較，此外還利用數位相機的像數的演進來做為實際檢測磨行事用性的依據。結果發現費雪模型在資料點多且可達到百分之百的滲透率或替代率時有較好的預測性，而甘伯茲模型則在當可以適當地找出適合的上限時有較好的表現。至於廣泛的羅吉斯模型不管在預測或者是模型的配適上都有不錯的表現。

The purpose of this thesis is to find a new technological forecasting model based on time-varied capacity and it has a flexible point of inflection. Technological forecasting is a good auxiliary tool to help managers make a decision whether in technological substitution or technological growth. It can also be used to find out the time that will reach the maximum growth rate and to know the possible trend of product’s life cycle. Consequently, a good forecasting can offer useful information to know the possible circumstances in the future. Fisher-Pry and Gompertz model are the most commonly used in this field, and many adapted models also based on their structure. Therefore, this thesis compares these models according to curves and mathematical criteria. The penetration of durable goods is used to test these three models. The result shows that Fisher-Pry model can fit the data well in some long data sets which reached 100% of capacity and Gompertz model fits data well when the right capacity has been set and data set is asymmetric. The extended logistic model shows a better fit and forecast in most data sets.