林心慧余英嘉2025-08-282025-08-282026-07-31U0061-2307202108401600https://www.airitilibrary.com/Publication/alDetailedMesh1?DocID=U0061-2307202108401600https://nutcir-lib.nutc.edu.tw/handle/123456789/1276在數位時代發展下，平台經濟興起加上COVID-19疫情影響，外送平台服務成為疫情期間新商機，餐廳面對民眾減少出門消費，消費者減少外食的頻率，讓餐飲業績受到衝擊使餐飲外送平台使用率提高。 本研究將運用 Python技術蒐集台灣兩大外送平台Foodpanda和Uber eats的大數據評論，首先彙整出消費者對餐廳、外送員、平台評論的平均分數、總分及意見字詞，藉此找出評論中的主要意見表達者進行情緒分析，其次，根據情緒分析結果設計問卷，問項為行為構面、生活型態構面、評論及人口統計構面，最後運用 SPSSModeler 中決策樹進行顧客分群，依變數為再購意願、向親友推薦及正向口碑傳遞，自變數為生活型態、評論、人口統計，依據消費者的特性分為高意願、中意願及低意願。 分析結果顯示，情緒分析在評論中Foodpanda消費者較在意餐廳中餐點品質，Uber eats消費者較在意平台上的系統品質及資訊品質。決策樹分析顯示，大部分消費者的再購意願、向親友推薦皆有高度意願，傳遞正向口碑意願較低，平均消費頻率為2~3周一次，平均每人花費為101~200元。本研究將根據情緒分析和決策樹分析結果，對餐飲外送平台和餐廳業者提出精準服務及目標行銷策略上之建議。In the development of the digital age, the rise of the platform economy and the impact of the COVID-19 epidemic, delivery platform services have become a new business opportunity during the epidemic. When restaurants face the public, they reduce their out-of-home consumption, consumers are more aware of food safety, and consumers reduce the frequency of eating out. So that the performance of the catering industry is affected, and the utilization rate of the catering delivery platform is increased. This research will use Python web crawler technology to collect big data reviews of Taiwan’s two major delivery platforms Foodpanda and Uber eats. First, the average score, total score and opinion words of consumers’ reviews of restaurants, delivery staff, and platforms will be collected. To find out the main opinion expressers in the comments for sentiment analysis. Secondly, design a questionnaire based on the results of sentiment analysis. The questions are behavioral aspects, Activities,Interests and Opinions Scale, comments and demographic aspects, and finally use SPSS Modeler to make decisions Tree, operating customer grouping, depending on the variables as repurchase willingness, recommendation to relatives and friends and positive word-of-mouth transmission, independent variables as Activities,Interests and Opinions Scale, reviews, demographics, and divided into high willingness, medium willingness and low willingness according to the characteristics of consumers. The results of the analysis show that in the reviews of sentiment analysis, Foodpanda consumers are more concerned about the quality of Chinese food in restaurants, while Uber eats consumers are more concerned about the quality of the system and information on the platform. Decision tree analysis shows that most consumers have high willingness to repurchase and recommend to relatives and friends, and low willingness to pass positive word-of-mouth. The average consumption frequency is 2-3 weeks, and the average spending per person is 101-200 yuan. Based on the results of sentiment analysis and decision tree analysis, this research will propose precise service and targeted marketing strategies to catering delivery platforms and restaurant operators.zh餐飲外送平台口碑大數據情緒分析決策樹Food delivery appsWord-of-mouthSentiment AnalyticsDecision treemaster thesis