Method and system for intelligent replenishment based on SKU

Abstract

The present invention relates to a technical solution for a SKU-based intelligent replenishment method and system, comprising: constructing a sample set based on raw data; building a prediction model based on time series decomposition, using the sample set and promotional activity information as inputs to the sales forecasting model, and integrating a custom moving average algorithm to output sales forecasts for the next cycle; and predicting the replenishment volume for the next cycle based on the sales forecasts and inventory turnover rate as influencing factors. This solution improves product replenishment efficiency while ensuring that the inventory turnover rate of the entire warehouse is within a safe range, making inventory and transportation costs more reasonable.

Claims

1. A method for intelligent replenishment based on SKU, characterized in that: The following steps are involved: S100, constructing a sample set based on the original data, including obtaining historical sales data, constructing a sample set based on stock keeping units, and a promotional sample set; Obtaining historical sales data and building sample sets based on SKUs and promotional sample sets includes: S110, setting condition information to filter out the historical sales data that meets the condition, wherein the condition information includes a historical time range, a store, and a stock keeping unit; S120: Set a time window for a period, use the time window as a period to aggregate sales within the period, and use the time information of the period and the aggregate sales value within the period to form a sample based on the stock keeping unit (SKU). Multiple samples within a historical time range constitute the sample set based on the SKU. S130, the times corresponding to the promotional events within the historical time range and the period to be predicted and the scale of the promotion constitute the promotion sample set based on the stock keeping unit; S200: Construct a prediction model based on time series decomposition, use the sample set as input to the sales forecast model, and integrate a custom moving average algorithm to output a sales forecast value for the next period. The integration of the custom moving average algorithm to output a sales forecast value for the next period includes: Modifying the sales forecast value for the next period based on the predicted growth rate, wherein the predicted growth rate is the relative growth rate between the window average and the sales forecast value for the next period; The sales forecast value for the next period is corrected based on the forecast growth rate, wherein the forecast growth rate is the relative growth rate between the window average and the sales forecast value for the next period, and includes: S221, setting a window for average value calculation, excluding a period with the maximum sales value and a period with the minimum sales value in the most recent window, and calculating the window average value for the remaining periods; S222, calculating the predicted growth rate, wherein the predicted growth rate is the sales forecast value of the next cycle divided by the window average value; S223, for a non-promotional day, if the predicted growth rate is outside the ideal predicted growth rate range, the sales forecast value for the next period is revised, and the revised sales forecast value for the next period is the window average value; S300 , predicting the replenishment quantity for the next cycle based on the sales forecast value for the next cycle and the inventory turnover rate as influencing factors.

2. The method of intelligent replenishment based on SKU according to claim 1, characterized in that: The method of predicting the replenishment quantity for the next cycle based on the sales forecast value and inventory turnover rate for the next cycle as influencing factors includes: Based on the sales forecast value of the next cycle, the inventory turnover rate of the previous cycle and the basic inventory turnover rate as influencing factors, the replenishment quantity of the next cycle is predicted.

3. The method of intelligent replenishment based on SKU according to claim 2, characterized in that: The forecasting of the replenishment quantity for the next cycle based on the sales forecast value for the next cycle, the inventory turnover rate of the previous cycle, and the basic inventory turnover rate as influencing factors includes: S310: The inventory turnover rate of the previous cycle is the accumulated inventory volume of the previous cycle divided by the accumulated outbound volume within the previous cycle; S320, setting the inventory turnover rate of the basic cycle; S330, the replenishment quantity forecast value of the next cycle is proportional to the sales forecast value of the next cycle, and inversely proportional to the ratio of the inventory turnover rate of the previous cycle to the inventory turnover rate multiple of the basic cycle.

4. A system based on SKU intelligent replenishment, characterized in that: Used to execute the SKU-based intelligent replenishment method according to any one of claims 1 to 3, comprising a raw data collection module, a historical sales data acquisition module, a sample set construction module, a sales forecast model construction module, a sales forecast value correction module, and a replenishment quantity prediction module; Among them, the original data collection module is used to collect order details, shipments, and inventory data tables, summarize the data tables, and count the sales, outbound quantity, and inventory information of each inventory unit every day; the historical sales data acquisition module is used to use the historical time range, store, and inventory unit as conditional information to filter out historical sales data that meets the conditions; the sample set construction module is used to construct the sample set required for model training and prediction, and construct samples of different formats based on historical sales data; the sales forecast model construction module is used to fit different factors and accumulate the factors to obtain a time series decomposition forecast model; the sales forecast value correction module is used to customize the moving average to correct the sales forecast value of the next cycle; the replenishment quantity prediction module is used to calculate the replenishment quantity forecast value of the next cycle based on the inventory turnover rate, basic inventory turnover rate and sales forecast value of the next cycle as influencing factors.

5. The SKU-based intelligent replenishment system according to claim 4 is characterized in that: It also includes forecast plan configuration module, forecast and evaluation indicator display module; Among them, the forecast plan configuration module is used to add a forecast plan, including setting the store name, inventory unit information, and generating records to record the forecast plan; the forecast and evaluation index display module is used to display the forecast indicators and evaluation indicators of the inventory units involved in each forecast plan.

Images