电商库存断货预警模型数据

电商断货预警算法
随着电商渠道竞争日益激烈。在发货时效服务方面，近几年逐步缩短为48、24、12小时发货时效， 平台对商家在物流发货时效指标方面的考核和罚款日益严格，同时直播电商背景下对供应链响应速度的极高要求。
本模型算法，针对零售行业尤其电商产业中供应链库存数字化管理与商业智能方面，在自有供应链的情况下，在多品种小批量多批次的供求背景下，
由于不同渠道在市场销售进度不一情况下， 实现断货的智能化提前预警干预与最优化库存水位管理。避免由于断货发货延迟导致的大量赔付损失以及库存备货过多导致的库存滞销压货风险，具有显著的意义。
特别突出的，本模型自研了5级BOM嵌套下的MRP运算模型，引入了根据物料计算出的物料对应成品的最小库存可维持天数， 进一步实现物料及成品库存的精准化估算库存水位，得到最佳采购点及断货预警触发点。
目前数据报表于24年6月上线艺福堂 -茶仓ERP - 产品状态变更 - 即将断货列表模块， 实现业务便捷化的断货预警管理。此外，艺福堂茶仓订单交付系统读取每日更新的安全库存数值， 提前触发成品及原料的断货预警，最后通过艺福堂物料询单系统，进行提前采购物料的催到付。断货预警模型主要是进行预测和识别即将断货的商品预警，
提前对销售端进行预警或者对采购生产进行加急干预。生产方式主要通过python代码，读取数仓中
基础数据， 然后进行报表运算，自动更新入库到艺福堂阿里云数据仓库，报表每日更新在1000条左右。
具体算法如下:1.根据成品库存 stock和商品近30日月销量 30day_sale，计算出库销比指标 stock_sale_ration， 即成品可维持天数。
2.通过金蝶BOM物料清单， 进行5级嵌套进行MRP物料需求计算，得到底层物料和其与成品的最终比例系数，算出每个物料对应成品可生产量 X1,X2,···Xi,
取某个成品对应底层物料的最小可生产数的最小值min(X1,X2,···Xi)， 最终获取该产品最小可生产量，与月销进行比率指标计算得到最小可生产量维持天数。
3.在库销比小于某个阈值同时最小可生产量维持天数小于某个阈值的情况下，触发成品断货预警。
4.通过商品销售、库存、原料、采购、销售量级、商品属性等维度特征，进行参数调整， 动态精准的对产品安全库存最终值调整，
最终进行安全库存水位的提前库存预警触发。

E-Commerce Stockout Warning Algorithm
As competition in e-commerce channels intensifies, delivery time requirements have gradually shortened to 48, 24, and 12 hours in recent years. Platforms have become increasingly strict in assessing and penalizing merchants for their logistics delivery timeliness, while the live-stream e-commerce context imposes extremely high demands on supply chain response speed.

This model algorithm targets digital supply chain inventory management and business intelligence in the retail industry, particularly the e-commerce sector. Under scenarios of in-house supply chains, demand-supply patterns featuring multiple varieties, small batches, and multiple batches, and inconsistent sales progress across different channels, it realizes intelligent early warning intervention for stockouts and optimal inventory level management. It effectively avoids substantial compensation losses caused by stockout-induced delivery delays and the risk of inventory stagnation and overstocking resulting from excessive stock preparation, thus holding significant practical value.

Notably, this model independently develops a Material Requirements Planning (MRP) calculation model under 5-level BOM (Bill of Materials) nesting, and introduces the minimum maintainable days of finished products corresponding to materials calculated based on the materials themselves. This further enables accurate estimation of inventory levels for both materials and finished products, and identifies the optimal procurement point and stockout warning trigger point.

Currently, the data report was launched in June 2024 on the Yifutang - Tea Warehouse ERP - Product Status Change - Upcoming Stockout List module, enabling convenient business-oriented stockout warning management. In addition, the Yifutang Tea Warehouse Order Delivery System reads the daily updated safety stock values to trigger early stockout warnings for finished products and raw materials, and finally, via the Yifutang Material Inquiry System, follows up on pre-procured materials for payment upon delivery and timely arrival. The stockout warning model mainly predicts and identifies early warnings for upcoming stockout commodities, and issues early warnings to the sales department or carries out urgent intervention for procurement and production.

The model is mainly implemented via Python code, which reads basic data from the data warehouse, performs report calculations, and automatically updates and imports the results into the Yifutang Alibaba Cloud Data Warehouse. The report is updated with approximately 1,000 entries per day.

The specific algorithm is as follows:
1. Calculate the stock-to-sales ratio indicator (stock_sale_ratio), which refers to the number of days the finished products can maintain sales, based on the finished product inventory (stock) and the 30-day average monthly sales volume (30day_sale) of the commodity.
2. Perform 5-level nested MRP material requirement calculation based on the Kingdee BOM list to obtain the final proportional coefficients between underlying materials and finished products. Calculate the producible quantity of finished products corresponding to each material as X1, X2, ···, Xi. Take the minimum value of the minimum producible quantities corresponding to the underlying materials of a certain finished product, i.e., min(X1,X2,···Xi), to obtain the minimum producible quantity of the product. Then calculate the maintainable days of the minimum producible quantity by comparing it with the monthly sales volume.
3. Trigger a finished product stockout warning when both the stock-to-sales ratio and the maintainable days of the minimum producible quantity are below their respective threshold values.
4. Dynamically and accurately adjust the final safety stock value of products based on dimensional features such as commodity sales, inventory, raw materials, procurement, sales volume tier, and commodity attributes, and finally trigger early inventory warnings based on the safety stock level.

搜集汇总

数据集介绍

特点

该数据集为电商库存断货预警模型数据，包含884条记录，每日更新，用于电商断货预警算法和库存管理。