Method and device for data prediction

Abstract

The present application discloses a method and device for data prediction. It relates to the field of computer information processing, and the method includes: obtaining order data within a predetermined time period, the order data including relevant information of commodities; performing aggregation processing on the order data to extract aggregated index data; inputting the aggregated index data into forecasting model to obtain the rating of the product; and predict the sales of the product based on the rating. The method and device for data prediction disclosed in this application can predict and control the quantity and inventory cost of commodities in the warehouse, and on this basis, maximize the timeliness of commodity delivery and improve customer satisfaction.

Description

technical field [0001] The present invention relates to the field of computer information processing, in particular to a method and device for data prediction. Background technique [0002] Supply chain management covers all aspects of product flow, information flow, and capital flow, and its goal is to minimize logistics and inventory costs while meeting customer needs. In the e-commerce industry, one of the keys to improving customer satisfaction is to reduce delivery time. Therefore, e-commerce companies will choose to build their own warehouses closer to customers so that they can have the shortest delivery distance and Fastest results. However, in reality, there is only a small amount of geographical space near the customers, resulting in a small warehouse that cannot store all SKUs (Stock keeping unit, the smallest unit of goods), so how to choose a part of SKUs that can accommodate It is the problem solved by this patent that it is stored in a warehouse close to the...

AI Technical Summary

Problems solved by technology

However, the actual business is more complicated. According to the above description of the existing technology, it can be found that the existing technology has the following four disadvantages: First, when customers place an order, they often purchase multiple SKUs at the same time. In the case of high-volume SKUs, in this case, the customer's order still cannot be fully shipped by the warehouse adjacent to the customer, resulting in an increase in fulfillment effectiveness

If the rules of whitelisting high-volume SKUs are followed, SKU1 will enter the finalized whitelist, but SKU2 will not, but in this case, only two customer orders can be fulfilled, and 60 customer satisfactions are lost Spend

Third, the existing technology is based on an assumption: the historical sales order of each SKU is consistent with the future sales order, that is, the SKU with the largest sales volume in the history and the top ranking will also have the largest sales volume in future orders

However, in reality, due to reasons such as promotions and seasons, history and the future are not completely consistent, so there is a large deviation in the existing technology

Fourth, based on the assumption that the history is different from the future, it can be considered that there are many factors that affect the future order volume of each SKU, while the existing technology only considers the information of the historical sales volume of the SKU

Images