後疫情時代之美國電影臺灣票房預測與分析

Abstract

電影欣賞是日常中極為普遍的娛樂方式，然而，隨著新冠肺炎疫情在世界各地爆發，人們的生活及消費模式改變，電影市場亦受到強烈衝擊，全球票房大幅衰退，若能提前預測電影票房，便能讓製片公司、發行商和投資者更容易決策。因此，本研究探討疫情爆發前後兩年(2018～2021)臺灣電影市場發展與變化。 本研究樣本為2018~2021年於臺灣上映之美國電影374部，結合電影基本資料、競爭關係、經濟因素與疫情因素預測電影票房。採用留一交叉驗證評估，並應用sci-kit learn中的線性迴歸、支持向量迴歸、決策樹與隨機森林迴歸，四種預測技術建立電影票房預測模型，評估指標包含R平方、平均絕對誤差與均方根誤差，結果顯示隨機森林迴歸預測效果最好，但疫情後較難預測，此外，本研究進一步探討疫情前後影響電影票房的特徵重要性分析。 結果顯示，疫情前預算特徵為最重要因素，疫情後首週放映院數為最重要之因素，預算特徵在疫情期間影響力大幅下降，動作類型環境因子影響力上升，意即動作類型電影在疫情期間競爭關係相對增加，此外，娛樂稅資料與疫情確診人數及死亡人數具一定程度關聯，疫情後娛樂稅之電影占比影響力上升，亦可代表電影票房受疫情環境影響。

Movie appreciation is an extremely common form of entertainment in daily life. However, as the COVID-19 epidemic breaks out around the world, people''s lives and consumption patterns have changed, and the movie market has also been strongly impacted. The global box office has declined sharply. If the movie box office can be predicted in advance, This will make decision-making easier for production companies, distributors and investors. Therefore, this study explores the development and changes of the Taiwan film market in the two years before and after the outbreak of the epidemic (2018~2021). The sample of this study is 374 American movies released in Taiwan from 2018 to 2021. The movie box office is predicted based on basic movie information, competitive relationships, economic factors and epidemic factors. Use leave-one-out cross-validation evaluation, and apply linear regression, support vector regression, decision tree and random forest regression in sci-kit learn. Four prediction technologies are used to establish a movie box office prediction model. The evaluation indicators include R square, mean absolute error and root mean square error, the results show that random forest regression has the best prediction effect, but it is more difficult to predict after the epidemic. In addition, this study further explores the importance analysis of features that affect movie box office before and after the outbreak of epidemic. The results show that the budget characteristics before the epidemic are the most important factor, and the number of theaters in the first week after the epidemic is the most important factor. The influence of budget characteristics dropped significantly during the epidemic, and the influence of action type environmental factors increased, which means that the impact of action movies during the epidemic period competition has increased relatively. In addition, entertainment tax data has a certain degree of correlation with the number of confirmed cases and deaths of the epidemic. After the epidemic, the influence of the entertainment tax on movies has increased, which can also mean that movie box office is affected by the epidemic environment.