串流影音服務之創新擴散模型研究：以Netflix和Spotify為例

為了探討串流影音服務業的創新擴散與預測，本研究先透過文獻探討選取了Bass Model、Gompertz Model及將網路使用人數作為基礎擴散的Contingent Diffusion Model等做為主要研究模型，本研究的主要目的是想探討影音串流平台之擴散與用戶數的成長是否適用創新擴散模型來解釋，亦想了解影視串流和音樂串流服務是否有不同的創新擴散模式，此外，本研究也針對過去學者對於最適解釋模型和最佳預測模型之質疑來進行研究上的確認，探討影音串流平台所適用之解釋模型和預測模型是否如先前學者之結論確實有差異。在檢測模型之解釋能力方面，本研究是透過判定係數和修訂Theil不等係數來處理，至於檢測模型預測能力方面，則使用MAPE值來衡量。本研究最後選擇的研究對象是目前發展最為快速，也累積最多用戶數的兩個影音串流平台—Netflix和Spotify。實證研究的結果顯示，影音串流平台確實可以透過創新擴散模型來觀察其成長趨勢，此外，不同於過去學者針對耐久財所得到之結論，本研究發現，影音串流平台的最適解釋模型和最佳預測模型是有一致性的，此意味著模型的解釋能力和預測能力具有相對的關係；最後，本研究亦發現串流平台中的不同服務類型(影視vs音樂)之差異並不會影響模型的解釋和預測結果。

This study explored the Bass Model, Gompertz Model, and Contingent Diffusion Model which uses the number of users on the Internet as a basis for research, to explore whether the user growth of the video streaming platform is applicable to the innovation diffusion model. Through observing number of subscribers of Netflix and Spotify, the two fast-growing video streaming platforms, the study discussed whether video streaming and music streaming services has different results on the innovation diffusion models. Besides, according to the query of scholars, the research also discussed whether explanatory and forecast model is the same model. It took the coefficient of determination and the Theil inequality coefficient to determine the goodness-of-fit and used the Mean Absolute Percentage Error to measure forecasting performance of these models. In the light of the empirical research, the study considered that video streaming platforms can be observed through the innovation diffusion model, moreover, it found the fittest explanatory model and the best forecast model of video streaming platforms are the same, no matter the result is from Netflix or Spotify. It indicated that the goodness-of-fit and forecasting ability of models is relevant.