Intelligent replenishment method, system and device based on end-to-end learning, and storage medium

Abstract

The invention discloses an intelligent replenishment method based on end-to-end learning. The method comprises the following steps: S1, acquiring data required by a mathematical model; S2, generatingsales features related to the sales volume according to the obtained data; S3, taking the characteristics predicted by the sales volume as original input values; establishing a mathematical model according to the total operation cost; the total operation cost being used as a loss function; solving and directly outputting suggest replenishment amount. Different from reduction of manual preprocessing and follow-up processing, direct output from original input to final output and direct output of results after replenishment by taking sales prediction features as original input values, end-to-endalgorithm learning is realized, so that the algorithm efficiency is improved, and the optimization space is expanded. The invention further provides a storage medium. The invention further provides anintelligent replenishment system. The invention further provides an intelligent replenishment device.

Description

【Technical field】 [0001] The present invention relates to the field of data prediction, in particular to an intelligent replenishment method, storage medium, system and device based on end-to-end learning. 【Background technique】 [0002] Forecasting is an important part of the operation process, by studying the past, understanding the present, and predicting the future, so as to maximize the benefits. [0003] Replenishment forecasting is one of them, and reasonable arrangements are made through supplementary testing; in the quantitative forecasting technology related to replenishment, goods are purchased and replenished by intelligently predicting sales volume, but the existing intelligent replenishment The delivery method is divided into two modules for implementation, generally sales forecast and replenishment decision. Sales forecast and replenishment decision are two sequential independent modules. The sales forecast result is input into the replenishment module, and th...

AI Technical Summary

Problems solved by technology

[0005] 1. The optimization goals of the two modules are different: the sales forecast aims to minimize the mean square error, and the replenishment takes business indicators as the goal, such as: profit maximization or cost minimization; due to the inconsistency between the forecast goal and the final decision goal, resulting in The result given by the forecast does not necessarily get the optimal value of the replenishment, ignoring the potential space for optimization, and the error of the sales forecast result may also lead to inaccurate replenishment

[0006] 2. The accuracy of sales forecast due to lack of data and other reasons is not enough, which affects the decision-making on the optimal recommended replenishment quantity

Images