http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Kaito Oshiro,Takeo Okazaki 대한전자공학회 2018 IEIE Transactions on Smart Processing & Computing Vol.7 No.1
This research forecasts order trends for drugstores based on order history datasets. Frequency–monetary–distance (FMD) analysis, merchandise clustering, and a modified model selection criterion are used to improve forecast accuracy. FMD analysis is a store classification method that is used for considering accessibility to the store as one of its characteristics. Merchandise clustering uses a wavelet transform aimed at finding the common order trends in products from the order history. The modified model selection criterion is intended to provide forecast models that reduce either underestimated or overestimated errors. Our verification results show that FMD analysis provides forecasts with fewer errors for most products, compared to the classification method without considering accessibility. Wavelet-based clustering reduces forecast errors depending on mother wavelets. The modified model selection criterion also reduces underestimated or overestimated errors.