在疫情影響或是科技的演變,電子商務已是社會環境中的主要趨勢,而此趨勢使得新創電商平台愈來愈多,對此,管控成本、賺取利潤、永續發展是主要的三大目標。因此,本研究以個案「一抹丘」之電商平台的角度,提出對銷售額影響之重要因子與準確度較高之預測方法,以其更了解未來之發展及適用之行銷策略。本研究利用人資料進行複迴歸模型與隨機森林預測模型之建置,從研究結果發現,在複迴歸模型中不顯著之變數,在隨機森林模型中,重要性雖亦不高,但對消費者之消費預測仍有影響;另外,本研究再藉由日資料進行時間序列模型與隨機森林模型之預測,從研究結果發現,隨機森林之預測準確度較時間序列模型佳,且利用轉折點法發現,隨機森林模型在預測銷售額之預測趨勢,能高度重合真實值,亦即具有高度的預測精確度。本研究以隨機森林法所建構之預測模型,可有效的提供電商平台進行倉儲管理、成本管控,不僅能使商品進貨更加精準,更能讓進貨成本得到管控,進而降低成本支出及庫存壓力。
Due to the impact of the epidemic and evolution of technology, e-commerce has become a major trend in the social environment, and this trend has led to more and more new e-commerce platforms. The controlling costs, earning profits, and sustainable development are the main three goals of the managing of e-commerce platforms. Therefore, this research focus on the "Atwhill" which is a new e-commerce platform, proposes important factors affecting sales and a forecasting method with higher accuracy, so as to better understand future development and applicable marketing strategies. This study used the data from the platforms to build different model. the results of multiple regression models and random forest prediction models showed, that variables not significant in the multiple regression model are not of high importance in the random forest model, but still have an impact on the consumption forecast of consumers. In addition, the results of time series and the random forest prediction models showed, that the prediction accuracy of the random forest is better than the time series model. Especially, in the turning point methods, this results showed, that the random forest model forecast trends in forecast sales, which can highly coincide with the true value, that is, it has a high degree of prediction accuracy. In the management view, the prediction model constructed by the random forest method in this study can effectively provide an e-commerce platform for warehouse management and cost control, which not only makes the purchase of goods more accurate, but also allows the cost of purchases to be controlled, thereby reducing cost expenditures and inventory pressure.