呂有豐Lue, Yeou-Feng黃至偉Huang, Chih-Wei2023-12-089999-12-312023-12-082022https://etds.lib.ntnu.edu.tw/thesis/detail/10cc298ea9bc2a1b70421a2535f29e1d/http://rportal.lib.ntnu.edu.tw/handle/20.500.12235/120708預測二手車的價格一直是消費者和汽車經銷商非常感興趣的話題。然而，由於預測二手車價需具備專業知識，且交易價格非常混亂，資訊也不透明，準確預測，並不容易。近年來，由於主要經濟體與開發中國家對汽車的需求暢旺，此外，由於中美貿易戰、新冠肺炎大流行、汽車晶片短缺、通貨膨脹等因素影響，新車供貨不足，導致二手車的需求量顯著提昇。價格為影響交易量的重要因素，能夠準確預測二手車價，對交易雙方極為重要。唯二手車價受製造年份、里程數、引擎排氣量、車型等諸多因素影響，準確預測車價格，並不容易，且少有研究機構或學者提出有效的解決方案。為解決此問題，本研究以基於優勢之約略集合法 (Dominance Based Rough Set Approach，DRSA) 推衍影響二手車價之核心屬性，作為預測二手車價的自變數，並以二次模糊分段迴歸模型，預測二手車價。為驗證分析架構之可行性，本研究自知名汽車拍賣網站8891.com.tw下載過去20年Toyota Altis、Honda CRV等六款車型之銷售紀錄3689筆，並以基於優勢之約略集合法擷取車齡、生產區域、品牌等七屬性為核心屬性，其後，以Toyota Altis 之銷售紀錄，預測車價。本究以20年資料中，前15年的資料訓練模型，並以後5年的資料實證模型之準確性。依據實證研究之結果，本研究所提出之模型，可準確預測二手車價。Predicting the price of second-hand cars has long been a topic of great interest to consumers and car dealers. However, it is not easy to predict the price of a second-hand car accurately because it requires expertise and transaction prices are chaotic and information is not transparent. In recent years, the demand for used cars has increased significantly due to the booming demand for automobiles in major economies and developing countries, as well as the shortage of new cars due to the trade war between China and the United States, the Coronavirus Disease 2019 (COVID-19) pandemic, the shortage of automotive chips, and inflation.Price is an important factor affecting the transaction volume. It is very important for both sides of the transaction to accurately predict the price of used cars. However, the price of second-hand cars is influenced by numerous factors such as the year of manufacturing, mileage, engine displacement and model. It is not easy to accurately predict the price of cars, and few research institutions or scholars have put forward effective solutions. To resolve this problem, this thesis uses the key attributes of Dominance Based Rough Set Approach (DRSA) as the independent variables for predictions, and then uses a quadratic fuzzy piecewise regression model to predict the prices of second-hand cars.To verify the feasibility of the analysis framework, this study downloaded 3689 sales records of six car models such as Toyota Altis and Honda CRV in the past 20 years from the famous auto auction website, 8891.com.tw, and extracted seven core attributes, such as car age, production region and brand, by using the DRSA. The sales record of Toyota Altis is used to predict the car price. The first 15 years of data were used to train the model, and the next 5 years were used to verify the accuracy of the model. According to the results of the empirical study, the model proposed in this study can accurately predict the price of used cars.二手車價預測基於優勢之約略集合法二次模糊分段迴歸模型used car priceforecastingDominance Based Rough Set Approach (DRSA)Quadratic Fuzzy Piecewise Regressionetd