COVID-19對台灣短期租賃旅宿業之衝擊與影響-Airbnb台北個案實證分析

2019年末，突如其來的疫情爆發，並迅速地在全球擴散，影響了眾多行業的生計，其中由於邊境封鎖、觀光客來台人數銳減，旅遊業受到了最直觀的衝擊，而即便在後疫情時代，國內旅遊的傾向不是過去熱門的城市，而是較為偏遠鄉下的地區，此現象讓都市的房東情況依然嚴峻。為了加以瞭解受到衝擊最嚴重的房源類型，本研究透過四種機器學習方式(支持向量機、梯度提升、隨機森林、類神經網路)分析來自Inside Aribnb網頁上的公開資料，並透過主成分分析達到降維並視覺化分析的結果，希望透過此研究能了解在疫情的衝擊下，對哪類的房源影響衝擊最大，做為以後類似情境如果再次發生，業者可以參考進行調整的依據。

關鍵字

機器學習； Airbnb ； COVID-19 ；共享經濟

並列摘要

Abstract: At the end of 2019, the sudden outbreak of the epidemic and its rapid spread across the world affected the livelihoods of many industries. Among them, due to border blockades and a sharp drop in the number of tourists, the tourism industry was the most intuitively impacted, even in the post-epidemic era.The tendency of domestic tourism is not the popular cities anymore as the past, but more the remote rural areas preferred. This phenomenon keeps the situation of urban landlords still severe. In order to understand the most severely impacted housing types, this research uses four machine learning methods (SVM, Random Forest, Neural Network, Gradient Boosting) to analyze public data from Inside Aribnb's webpage, and achieves through principal component analysis hopefully. The results of dimensionality reduction and visual analysis from this research, could help us understand which types of listings would survive under the impact of the epidemic. If similar situations occur again in the future, the industry can refer to the basis for adjustments.