台灣共享汽車使用頻率與區域興趣點之研究

共享經濟是一種創新的經濟模式，它透過技術平台實現資源的共享與交換，從而提高了資源利用效率並降低了消費者的成本。本研究旨在了解興趣點分布如何影響共享汽車的使用情況。首先，透過傳統的卜瓦松迴歸（PR）模型來進行變數篩選，共篩選出租車資訊（如性別、年齡、使用里程..等）、興趣點資訊（如景點、餐廳、住宿…等）及社會因素（如人口密度、所得..等）等三大類合計16個解釋變數後，建立卜瓦松迴歸模型，並透過Moran’s I檢驗模型存在空間相依性。其次，建立16個解釋變數對共享汽車使用頻率的地理加權卜瓦松迴歸（GWPR）模型後，檢驗模型變數的空間異質性發現，除住宿至該區中心的平均距離變數外，其餘15個變數均具空間異質性。接下來，以AIC與AICc來檢驗模型的優劣，結果發現GWPR的AIC與AICc均低於PR。最後，將具空間異質性的15個解釋變數繪製六個直轄市158個行政區的顯著性地圖，來說明分布情況。研究結果顯示，興趣點的分布對共享汽車的使用頻率具有顯著影響。在租車資訊上，男性和40歲以下用戶頻繁使用共享汽車，且短程租賃需求較高，休旅車因其多功能性受到偏愛；在興趣點資訊上，自然賞景活動和文化體驗活動的比例對租賃次數有負向影響；在社會因素上，人口密度、所得中位數與大專院校數量對租賃次數也有正向影響。

關鍵字

共享汽車；空間異質性；卜瓦松迴歸；地理加權卜瓦松迴歸

並列摘要

The sharing economy is an innovative model that uses technology to share resources, thereby increasing efficiency and reducing costs. This study examines how points of interest（POIs）affect the usage of shared cars. Firstly, a Poisson Regression（PR）model was employed to select sixteen variables across three categories: car rental information（e.g., gender, age, mileage）, POI details（e.g., tourist spots, restaurants）, and social factors（e.g., population density, income）. After establishing the PR model and confirming spatial dependency with Moran's I test. Secondly, Geographically Weighted Poisson Regression（GWPR）model was applied. It revealed spatial heterogeneity in all but one variable the average distance to the district center. Furthermore, according to the AIC and AICc, the GWPR model performed better than the PR model, indicating greater fit and efficiency. Finally, Significance maps for the fifteen spatially heterogeneous variables across 158 districts showed that the distribution of POIs significantly influences the frequency of shared car use. Rentals are more common among men and individuals under 40, who prefer short-term leases and SUVs due to their versatility. In contrast, scenic and cultural activities negatively impact rental frequencies, while higher population density, median income, and the presence of universities positively influence them.