新產品的發展是企業能否在當代市場中生存的關鍵,然而當一位管理者想開發一項新產品,或想從產品擴散的先進國引進一項跨國新產品時,卻會面臨釵h的挑戰。因此,在訂定新產品的研發、行銷策略之前,企業必須獲得的重要資訊,即該項新產品是否有足夠的市場潛量,以及它是否會擴散得夠快。 在過去,受限於多種無法量化的外部環境因素,預測產品生命週期與銷售量的科學模型難以設立。幸運的是近年來由於Bass擴散模式的發展以及運用高效率的統計分析,預測新產品生命週期與銷售量的模型日趨穩定。 本篇論文將以過去的研究為基礎,應用不同的估計方法與統計模式來建立預測模型,進一步將不同的、可能影響新產品擴散過程的因素導入模型。藉由不同的模型之比較,期待能找出一個預測能力較好的預測模型,協助企業訂定更有效的行銷策略。 由於消費者對於無形產品的偏好較有形產品更難被觀察與衡量,其擴散也就更難預測,因此本論文將採用在台上映之美國電影為測試模型的實證對象。
Product development is the key to success for a company in the modern business world. However, managers face enormous challenges when planning to develop a new product or introduce an international new product from a country who is a forerunner of the new product’s diffusion. Therefore, before setting the development and promotion strategies of a new product, the most important information companies have to find out is whether the new product has sufficient market potential, and whether it will diffuse fast enough. Due to the limitation of unquantifiable external environmental factors, in the past, it was difficult to set up a scientific method to predict the life cycle and sales of a new product. Fortunately, with the development of Bass Diffusion Model and the application of highly efficient statistics analysis over the past years, the model is now more stable in forecasting the life cycle and sales of new products. In this thesis, different estimation and statistics approaches will be applied to build forecasting models based on the previous research. Furthermore, introduce different factors which may influence the diffusion process of the new product into the models. By comparing different models, it is hoped to find an efficient forecasting model with greater forecasting ability and help companies in making more effective marketing strategies. Since customers’ preference to intangible products is more difficult to be observed and measured than tangible ones, it is more difficult to forecast the diffusion as well. Therefore, American movies released in Taiwan will be taken as the subject for practically testing the forecasting models in this thesis.