Replenishment method and device

Abstract

The present invention discloses a replenishment method and device, which relate to the field of warehouse management technology. A specific implementation of the method includes: obtaining the first historical sales data of the target item in the first time period; constructing target feature data based on the first historical sales data; using a pre-trained replenishment strategy classification model and the target feature data, determining the target replenishment strategy adapted for the target item in the second time period, and replenishing based on the target replenishment strategy; wherein the replenishment strategy classification model is trained based on the third historical sales data of the third time period, the fourth historical sales data of the fourth time period, the preset first replenishment strategy, and the preset second replenishment strategy. This implementation can automatically determine the replenishment strategy adapted to the target item, flexibly and specifically recommend the most suitable replenishment strategy, improve the replenishment effect, and reduce inventory costs.

Description

Technical Field

[0001] The present invention relates to the technical field of warehouse management, and in particular to a replenishment method and device. Background Art

[0002] Currently, there are a large number of different inventory replenishment strategies and models, such as the common RSS strategy, as well as quantile replenishment and end-to-end replenishment strategies generated by applying new technologies and theories. Different replenishment strategies have different applicable scenarios and product characteristics. If all products use the same strategy at different times, or blindly choose from existing strategies, it will lead to poor replenishment results, resulting in out-of-stock costs, or inventory costs due to over-replenishment. Therefore, it is very necessary to flexibly recommend the most suitable strategy for different merchants and different products at different times (such as different seasons and promotional periods) to improve replenishment results and reduce various costs. Summary of the invention

[0003] In view of this, an embodiment of the present invention provides a replenishment method and device, which can automatically determine a replenishment strategy suitable for a target item, dynamically recommend the most suitable replenishment strategy in a targeted manner, improve replenishment effects, and reduce inventory costs.

[0004] To achieve the above objective, according to one aspect of an embodiment of the present invention, a replenishment method is provided, comprising:

[0005] Obtain the first historical sales data of the target item in the first time period;

[0006] constructing target feature data based on the first historical sales data;

[0007] Using the pre-trained replenishment strategy classification model and the target feature data, determine the target replenishment strategy adapted for the target item in the second time period, and replenish based on the target replenishment strategy; wherein the replenishment strategy classification model is trained based on the third historical sales data of the third time period, the fourth historical sales data of the fourth time period, the preset first replenishment strategy and the preset second replenishment strategy, the third historical sales data of the third time period is used to construct sample feature data of the replenishment strategy classification model, the fourth historical sales data of the fourth time period is used to construct sample label data of the replenishment strategy classification model, the fourth time period is later than the third time period, the second time period is later than the first time period, and the first time period is later than the fourth time period or is the same time period as the fourth time period.

[0008] Optionally, the method further comprises constructing sample label data of the replenishment strategy classification model according to the following process:

[0009] The fourth historical sales data is simulated using a preset first replenishment strategy and a preset second replenishment strategy respectively to obtain a first simulated inventory index and a second simulated inventory index; and sample label data is constructed based on the first simulated inventory index and the second simulated inventory index.

[0010] Optionally, the sample characteristic data includes statistical characteristic data and / or time series characteristic data.

[0011] Optionally, the statistical feature data includes one or more of the following: historical sales mean, first quantile value, median, standard deviation and maximum value.

[0012] Optionally, the time series characteristic data includes one or more of the following: autocorrelation coefficient, partial correlation coefficient, first-order difference, multi-order difference, coefficient of variation and average value of the number of days between non-zero sales volumes.

[0013] Optionally, when the preset first replenishment strategy or the preset second replenishment strategy is a risk value replenishment strategy, the sample feature data further includes attribute feature data, wherein the attribute feature data is determined according to the following process:

[0014] Based on the third historical sales data, determine the second quantile value and the average value of the third historical sales data;

[0015] The attribute feature data is calculated according to the second quantile value and the average value of the third historical sales data.

[0016] Optionally, the first simulated inventory index includes a first spot rate and a first inventory turnover days; the second simulated inventory index includes a second spot rate and a second inventory turnover days;

[0017] Based on the first simulation inventory index and the second simulation inventory index, constructing sample label data includes:

[0018] Comparing the first spot rate and the second spot rate, the first inventory turnover days and the second inventory turnover days to obtain a first result;

[0019] Based on the first result, the sample label data is constructed.

[0020] Optionally, the first simulated inventory indicator includes a first target inventory; the second simulated inventory indicator includes a second target inventory;

[0021] Based on the first simulation inventory index and the second simulation inventory index, constructing sample label data includes:

[0022] Determining a first real sales volume corresponding to the first target inventory and a second real sales volume corresponding to the second target inventory;

[0023] respectively determining a first difference between the first target inventory and the first actual sales volume, and a second difference between the second target inventory and the second actual sales volume;

[0024] The sample label data is constructed according to the first difference and the second difference.

[0025] Optionally, constructing the sample label data according to the first difference and the second difference includes:

[0026] Determine the positive and negative values ​​of the first difference and the second difference respectively; wherein, if the first target inventory is greater than or equal to the first actual sales volume, the first difference is a positive error, otherwise the first difference is a negative error, and if the second target inventory is greater than or equal to the second actual sales volume, the second difference is a positive error, otherwise the second difference is a negative error;

[0027] The sample label data is constructed according to the positive and negative values ​​of the first difference and the second difference.

[0028] To achieve the above object, according to another aspect of an embodiment of the present invention, a replenishment device is provided, comprising:

[0029] A data acquisition module, used to acquire first historical sales data of a target item in a first time period;

[0030] A feature construction module, used for constructing target feature data based on the first historical sales data;

[0031] a replenishment determination module, configured to determine a target replenishment strategy adapted for the target item in the second time period by using a pre-trained replenishment strategy classification model and the target feature data, and to perform replenishment based on the target replenishment strategy; wherein the replenishment strategy classification model is trained based on the third historical sales data of the third time period, the fourth historical sales data of the fourth time period, the preset first replenishment strategy, and the preset second replenishment strategy; the third historical sales data of the third time period is used to construct sample feature data of the replenishment strategy classification model; the fourth historical sales data of the fourth time period is used to construct sample label data of the replenishment strategy classification model; the fourth time period is later than the third time period, the second time period is later than the first time period, and the first time period is later than the fourth time period or is the same time period as the fourth time period.

[0032] Optionally, the device also includes a model training module, which is used to: simulate the fourth historical sales data using a preset first replenishment strategy and a preset second replenishment strategy, respectively, to obtain a first simulated inventory indicator and a second simulated inventory indicator; and construct sample label data based on the first simulated inventory indicator and the second simulated inventory indicator.

[0033] Optionally, when the preset first replenishment strategy or the preset second replenishment strategy is a risk value replenishment strategy, the sample feature data further includes attribute feature data, and the model training module is further used to determine the attribute feature data according to the following process:

[0034] Based on the third historical sales data, determine the second quantile value and the average value of the third historical sales data;

[0035] The attribute feature data is calculated according to the second quantile value and the average value of the third historical sales data.

[0036] Optionally, the first simulated inventory index includes a first spot rate and a first inventory turnover days; the second simulated inventory index includes a second spot rate and a second inventory turnover days;

[0037] The model training module is also used to: compare the first spot rate and the second spot rate, the first inventory turnover days and the second inventory turnover days to obtain a first result; and construct the sample label data based on the first result.

[0038] Optionally, the first simulated inventory indicator includes a first target inventory; the second simulated inventory indicator includes a second target inventory;

[0039] The model training module is also used to: determine a first real sales volume corresponding to the first target inventory, and a second real sales volume corresponding to the second target inventory; respectively determine a first difference between the first target inventory and the first real sales volume, and a second difference between the second target inventory and the second real sales volume; and construct the sample label data based on the first difference and the second difference.

[0040] Optionally, the model training module is also used to: respectively determine the positive and negative values ​​of the first difference and the second difference; wherein, if the first target inventory is greater than or equal to the first actual sales volume, the first difference is a positive error, otherwise the first difference is a negative error, and if the second target inventory is greater than or equal to the second actual sales volume, the second difference is a positive error, otherwise the second difference is a negative error; construct the sample label data according to the positive and negative values ​​of the first difference and the second difference.

[0041] To achieve the above-mentioned purpose, according to another aspect of an embodiment of the present invention, there is provided an electronic device, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the replenishment method of an embodiment of the present invention.

[0042] To achieve the above objective, according to another aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored, and when the program is executed by a processor, the replenishment method of the embodiment of the present invention is implemented.

[0043] An embodiment of the above invention has the following advantages or beneficial effects: through a pre-trained replenishment strategy classification model and target feature data constructed based on the first historical sales data of the target item in the first time period, a target replenishment strategy adapted for the target item in the second time period is determined, and a technical means for determining the replenishment quantity of the target item in the second time period based on the target replenishment strategy can automatically determine the replenishment strategy adapted to the target item, and dynamically recommend the most suitable replenishment strategy in a targeted manner, thereby improving the replenishment effect and reducing inventory costs.

[0044] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0046] Figure 1 is a schematic diagram of the main process of training a replenishment strategy classification model in the replenishment method of an embodiment of the present invention;

[0047] Figure 2 is a schematic diagram of the main process of the replenishment method according to an embodiment of the present invention;

[0048] Figure 3 is a schematic diagram of the principle of the replenishment strategy in the replenishment method of an embodiment of the present invention;

[0049] Figure 4 is a schematic diagram of main modules of a replenishment device according to an embodiment of the present invention;

[0050] Figure 5 is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0051] Figure 6 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0053] At present, in practical applications, the application of various replenishment strategies in different scenarios is mainly based on practical experience. Even for the same merchant and product at different times, the appropriate replenishment strategies are different. However, at present, basically the same strategy is adopted, which lacks real-time performance, flexibility, and adjustability. In addition, the inventory management and policy configuration of massive materials will gradually exceed the scope of manual operation and screening capabilities, increase the difficulty of inventory management and policy configuration, and require automated recommendation management. In order to solve at least one of the above technical problems, an embodiment of the present invention provides a replenishment method, which dynamically determines the replenishment strategy suitable for different items at different times among a variety of different replenishment strategies, and replenishes based on the determined replenishment strategy.

[0054] The replenishment method of the embodiment of the present invention simulates the historical sales data of the same item under multiple (for example, two) replenishment strategies based on massive historical sales data to simulate the supply chain inventory management performance. The historical sales data and the indicator data output by the simulation are used as sample data, and the machine learning method is used to train the replenishment strategy classification model. According to the training model, the appropriate replenishment strategy for each item at different times in the future is determined, and replenishment is performed according to the determined replenishment strategy.

[0055] To facilitate understanding of the replenishment method according to the embodiment of the present invention, the training process of the replenishment strategy classification model is first described below.

[0056] Figure 1 FIG. 1 is a schematic diagram of the main process of training a replenishment strategy classification model in a replenishment method according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0057] Step S101: Acquire third historical sales data of a third time period and fourth historical sales data of a fourth time period, wherein the fourth time period is later than the third time period.

[0058] Among them, the historical sales data obtained in this step can be the historical sales data of all categories of goods, or the historical sales data of a certain category of goods. If the historical sales data obtained is the historical sales data of a certain category of goods, the model training can be performed for each category with the category as the granularity to obtain the model corresponding to each category. When predicting the replenishment of the target item in the future, the model corresponding to the category to which the target item belongs can be selected for prediction. The third time period and the fourth time period can be a few days or dozens of days, such as 30 days or 60 days. In practical applications, the third time period and the fourth time period can be flexibly set according to the scene, such as the set replenishment effect and the computing power of the server, and the present invention does not limit this. In this embodiment, the fourth historical sales data of the fourth time period is used to construct the sample label data, and the third historical sales data of the third time period is used to construct the sample feature data. The sample feature data is the input data of the model, and the sample label data is the output data of the model. Therefore, the fourth time period is later than the third time period. In this embodiment, the time interval of the fourth time period is consistent with that of the third time period, and the fourth time period is later than the third time period as a whole.

[0059] As an example, the third time period may be from March 1, 2020 to March 30, 2021, and the fourth time period may be from April 1, 2020 to April 30, 2020. In another example, the third time period may be from March 1, 2020 to April 30, 2021, and the fourth time period may be from May 1, 2020 to May 30, 2020. May 1, 2020 to May 30, 2020 is a future time relative to March 1, 2020 to April 30, 2021. By training the relationship between the sales data from March 1, 2020 to April 30, 2021 and the sales data from May 1, 2020 to May 30, 2020, a replenishment strategy classification model is obtained, and the replenishment strategy classification model can predict the optimal replenishment strategy to be adopted in the future. Then, the replenishment strategy classification model and the current sales data can be used to predict the replenishment strategy that needs to be selected for replenishment in the next time period.

[0060] In an optional embodiment, after obtaining the third historical sales data and the fourth historical sales data, data cleaning processing may be performed. For example, the values ​​of fields such as warehouse, category, name (sku), time, and sales volume in the sales data are extracted, and the NAN value is filled, for example, with 0 or the mean. Among them, the NAN value refers to Not A Number, which is a type of value of the numerical data type in computer science, indicating an undefined or unrepresentable value.

[0061] Step S102: constructing sample feature data based on the third historical sales data.

[0062] In this embodiment, the sample characteristic data may include statistical characteristic data and / or time series characteristic data.

[0063] Among them, the statistical characteristic data includes one or more of the following: historical sales mean, first quantile value, median, standard deviation and maximum value. Among them, the first quantile value can be flexibly set, such as the 90% quantile value or the 95% quantile value. Quantile, also known as quantile, refers to the numerical point that divides the probability distribution range of a random variable into several equal parts. Commonly used ones are median (i.e., quantile), quartile, percentile, etc. Quantile is a point in the continuous distribution function, and this point corresponds to probability p.

[0064] Time series feature data includes one or more of the following: autocorrelation coefficient, partial correlation coefficient, first-order difference, multi-order difference, coefficient of variation and average number of days between non-zero sales. Among them, the autocorrelation coefficient measures the correlation between the same event in two different periods, which is to say that it measures the impact of one's past behavior on one's present. The partial correlation coefficient means that in a system composed of multiple factors, when studying the influence or correlation of one factor on another factor, the influence of other factors is regarded as a constant (remains unchanged), that is, temporarily ignoring the influence of other factors, and studying the closeness of the relationship between the two factors separately. The numerical result is the partial correlation coefficient. The first-order difference refers to the difference between two consecutive adjacent terms in a discrete function. The second-order difference refers to the difference of the first-order difference, and the higher-order difference refers to the difference of a certain order difference. The coefficient of variation is calculated based on the standard deviation and average value of the historical sales data. Specifically, the coefficient of variation = (standard deviation / average value) × 100%. The average number of days between non-zero sales refers to the average number of days between any two non-zero sales.

[0065] In an optional embodiment, when the preset first replenishment strategy or the preset second replenishment strategy is a risk value replenishment strategy (VAR replenishment strategy, Value At Risk, referred to as VAR), since the risk value replenishment strategy is relatively more suitable for replenishing long-tail products, it is necessary to set attribute characteristics that reflect the long-tail situation of historical demand, that is, the sample feature data also needs to include attribute feature data. Specifically, the attribute feature data can be determined according to the following process:

[0066] Based on the third historical sales data, determine the second quantile value and the average value of the third historical sales data;

[0067] The attribute feature data is calculated based on the second quantile value and the average value of the third historical sales data. For example, the quotient obtained by dividing the second quantile value by the average value of the third historical sales data is used as the attribute feature data. More specifically, the quotient obtained by dividing the 95% quantile of the historical sales by the average value of the third historical sales data can be set as the attribute feature.

[0068] The embodiment of the present invention can improve the accuracy of the model through multi-dimensional feature data.

[0069] Step 103: Simulate the fourth historical sales data using a preset first replenishment strategy and a preset second replenishment strategy to obtain a first simulated inventory index and a second simulated inventory index; construct sample label data based on the first simulated inventory index and the second simulated inventory index.

[0070] As an example, the first replenishment strategy and the second replenishment strategy may be a VAR replenishment strategy and an RSS replenishment strategy. The RSS replenishment strategy (Reorder point Safety Stock, Reorder point refers to the point where replenishment is required, and Safety Stock refers to safe stock fluctuations) refers to a replenishment strategy that starts replenishment when the inventory drops to the replenishment point R and takes safety stock fluctuations into consideration.

[0071] For sample label data, the embodiment of the present invention provides two construction methods, which are as follows:

[0072] The first method: The first simulated inventory index includes the first spot rate and the first inventory turnover days; the second simulated inventory index includes the second spot rate and the second inventory turnover days. Among them, the spot rate refers to the proportion of days that the item meets the demand within a period of time. Specifically, the spot rate can be obtained by the proportion of days that the replenishment quantity (replenishment quantity equals the target inventory minus the current inventory) obtained by the replenishment strategy simulation is higher than the actual sales volume. For example, if the number of days that the replenishment quantity is higher than the actual sales volume is 21 days within 30 days, the spot rate is: 21÷30=70%. Inventory turnover days refers to the number of days required for the items in the inventory to be sold out in theory, which can be obtained based on the sum of the end-of-period inventory and the sum of sales. Specifically, the inventory turnover days can be equal to the sum of the end-of-period inventory divided by the sum of sales. For example, within 30 days, a certain item has replenishment and sales every day, and the sum of the end-of-period inventory is equal to the replenishment quantity minus the sales volume. The inventory turnover days of the item are equal to the sum of the end-of-period inventory of 30 days divided by the sum of the sales volume of 30 days.

[0073] After obtaining the first simulated inventory index and the second simulated inventory index, the indexes under the two replenishment strategies are compared according to the same item granularity. Specifically, the comparison includes:

[0074] Comparing the first spot rate and the second spot rate, the first inventory turnover days and the second inventory turnover days to obtain a first result;

[0075] Based on the first result, the sample label data is constructed.

[0076] Take the VAR replenishment strategy and the RSS replenishment strategy as examples. Take the RSS replenishment strategy as the benchmark inventory strategy. If the spot rate of the VAR replenishment strategy is greater than the spot rate of the RSS replenishment strategy, and the inventory turnover days of the VAR replenishment strategy is less than or equal to the inventory turnover days of the RSS replenishment strategy, then it is marked as 1, and the rest are marked as 0.

[0077] The above method of constructing label data is relatively intuitive, but because its construction logic is too strict, it will lead to the problem of unbalanced sample label data (that is, too few sample data with label 1).

[0078] The second method: the first simulation inventory indicator includes the first target inventory; the second simulation inventory indicator includes the second target inventory.

[0079] Take the VAR replenishment strategy and RSS replenishment strategy as examples. The principle of the VAR replenishment strategy: given a sample list (set or list), how to calculate the approximate VAR. Assuming we know the cumulative distribution, VAR is defined as VaR α (Z)=inf{z|FZ(z)≥α}. According to the idea of ​​VAR method, if the sample size is the historical sales volume, the probability density function F(D) of the historical sales volume D in a period of time is calculated, the confidence level α is set, and the value corresponding to the quantile α is obtained as VAR. (α) It can be interpreted as the possibility of α in the future, and sales will not exceed VAR one day (α) , so it can be used as the maximum order quantity. Specifically, the method for calculating the target inventory of the VAR replenishment strategy is as follows: According to the list of historical sales, fit the empirical distribution, and calculate the 95% quantile of the empirical distribution, which is the VAR (95％) , taking it as the target inventory for one day, the target inventory for VLT+NRT+BP day is:

[0080]

[0081] Among them, VLT (Vendor Lead Time) refers to the length of time it takes for a supplier to deliver goods, from the creation (order creation) to the completion (shelfing) of a purchase order. NRT (next run time) refers to the time until the next order date, which is equal to the difference between the next order date and the current date. BP (Buying Period): The stocking period refers to how many days of sales the purchase quantity is planned to meet. If you expect this purchase to meet 30 days of sales, the stocking period should be set to 30 days.

[0082] like Figure 3 As shown in the figure, the target inventory of the RSS replenishment strategy is calculated based on the sum of the forecast demand within VLT+NRT+BP days and the safety stock, as shown in the following formula:

[0083]

[0084] Where k is the service level, which takes values ​​in [0, 1]; z k μ is the quantile of demand distribution when the service level is k. VLT is the variance of VLT, δ D is the variance of historical demand, that is, the variance of historical sales data, μ D is the mean of the forecasted demand, δ VLT+NRT is the variance of VLT+NRT. In the RSS replenishment strategy, the safety stock is mainly used to better cope with the fluctuation of the forecast and prevent the forecast deviation from being too large.

[0085] After obtaining the first simulated inventory index and the second simulated inventory index, calculate the error between the target inventory and the actual sales volume obtained by the two replenishment strategies at the sku granularity (the two replenishment strategies calculate the target inventory of VLT+NRT+BP days, and the actual sales volume is also the actual sales volume of VLT+NRT+BP days). Then, according to the same sku granularity, compare the errors under the two replenishment strategies to obtain sample label data. Specifically, the process includes:

[0086] Determining a first real sales volume corresponding to the first target inventory and a second real sales volume corresponding to the second target inventory;

[0087] respectively determining a first difference between the first target inventory and the first actual sales volume, and a second difference between the second target inventory and the second actual sales volume;

[0088] The sample label data is constructed according to the first difference and the second difference.

[0089] More specifically, after obtaining the first difference and the second difference, the process further includes:

[0090] Determine the positive or negative nature of the first difference and the second difference respectively; wherein, if the first target inventory is greater than or equal to the first actual sales volume, the first difference is a positive error, otherwise the first difference is a negative error, and if the second target inventory is greater than or equal to the second actual sales volume, the second difference is a positive error, otherwise the second difference is a negative error;

[0091] Construct the sample label data according to the positive or negative nature of the first difference and the second difference.

[0092] For the same sku, if:

[0093] It is a positive error under the VAR strategy and a negative error under the RSS strategy, and in this case it is marked as 0;

[0094] It is a positive error under the VAR strategy and a positive error under the RSS strategy, and VAR(positive error) <= RSS(positive error), and in this case it is marked as 1;

[0095] It is a positive error under the VAR strategy and a positive error under the RSS strategy, and VAR(positive error) > RSS(positive error), and in this case it is marked as 0;

[0096] It is a negative error under the VAR strategy and a positive error under the RSS strategy, and in this case it is marked as 0;

[0097] It is a negative error under the VAR strategy and a negative error under the RSS strategy, and VAR(negative error) >= RSS(negative error), and in this case it is marked as 1;

[0098] It is a negative error under the VAR strategy and a negative error under the RSS strategy, and VAR(negative error) < RSS(negative error), and it is marked as 0.

[0099] The sample label data obtained by the above method for constructing label data is relatively balanced, but the dynamic characteristics of different time dimensions are not considered. Therefore, a third method is proposed, adding the time dimension, as follows.

[0100] Step S104: Based on the sample feature data and the sample label data, train to obtain the replenishment strategy classification model. In this embodiment, the sample feature data is the input data of the model, and the sample label data is the output data of the model. The purpose of training the replenishment strategy classification model is to learn the relationship between the sample feature data and the sample label data.

[0101] The model trained in the embodiment of the present invention is a classification model. For the above-mentioned VAR replenishment strategy and RSS replenishment strategy examples, the model is a binary classification model. Therefore, the embodiment of the present invention selects the Xgboost model (Extreme GradientBoosting) for training to obtain a replenishment strategy classification model.

[0102] After the replenishment strategy classification model is trained, test data can be obtained to test the effect of the replenishment strategy classification model using the test data, and the replenishment strategy classification model can be further optimized according to the test results. After the final replenishment strategy classification model is obtained, the replenishment strategy classification model can be used to predict the replenishment strategy adapted to the target item at a certain time.

[0103] Figure 2 FIG. 1 is a schematic diagram of the main steps of a replenishment method according to an embodiment of the present invention. Figure 2 As shown, the method includes:

[0104] Step S201: Obtain the first historical sales data of the target item in the first time period. The target item is an item for which replenishment is to be predicted. The first time period may be later than the fourth time period, or may be the same as the fourth time period. For example, the fourth time period is from May 1, 2021 to May 30, 2021. The first time period may be from May 1, 2021 to May 30, 2021 (in this case, the second time period may be from June 1, 2021 to June 30, 2021, i.e., the replenishment strategy for June is predicted using the sales data for May), or from June 1, 2021 to June 30, 2021 (in this case, the second time period may be from July 1, 2021 to July 30, 2021, i.e., the replenishment strategy for July is predicted using the sales data for June).

[0105] Step S202: construct target feature data based on the first historical sales data, wherein the target feature data may include statistical feature data and / or time series feature data.

[0106] The statistical characteristic data includes one or more of the following: historical sales mean, first quartile value, median, standard deviation, and maximum value.

[0107] The time series characteristic data includes one or more of the following: autocorrelation coefficient, partial correlation coefficient, first-order difference, multi-order difference, coefficient of variation, and average number of days between non-zero sales.

[0108] When the preset first replenishment strategy or the preset second replenishment strategy is a risk value replenishment strategy, the target feature data further includes attribute feature data, wherein the attribute feature data is determined according to the following process:

[0109] Based on the first historical sales data, determining a second quantile value and an average value of the first historical sales data;

[0110] The attribute feature data is calculated based on the second quantile value and the average value of the first historical sales data. For example, the quotient obtained by dividing the second quantile value by the average value of the first historical sales data is used as the attribute feature data. More specifically, the quotient obtained by dividing the 95% quantile of the first historical sales by the average value of the first historical sales data can be set as the attribute feature.

[0111] Step S203: using the pre-trained replenishment strategy classification model and the target feature data, determining a target replenishment strategy adapted for the target item in the second time period, and performing replenishment based on the target replenishment strategy.

[0112] The replenishment method of the embodiment of the present invention determines a target replenishment strategy adapted for the target item in the second time period through a pre-trained replenishment strategy classification model and target feature data constructed based on the first historical sales data of the target item in the first time period, and a technical means for determining the replenishment quantity of the target item in the second time period based on the target replenishment strategy. The method can automatically determine a replenishment strategy adapted for the target item, dynamically recommend the most suitable replenishment strategy in a targeted manner, improve replenishment effects, and reduce inventory costs.

[0113] Figure 4 is a schematic diagram of the main modules of a replenishment device 400 according to an embodiment of the present invention. Figure 4 As shown, the replenishment device 400 includes:

[0114] The data acquisition module 401 is used to acquire the first historical sales data of the target item in the first time period;

[0115] A feature construction module 402, configured to construct target feature data based on the first historical sales data;

[0116] The replenishment determination module 403 is used to determine the target replenishment strategy adapted for the target item in the second time period by using the pre-trained replenishment strategy classification model and the target feature data, and perform replenishment based on the target replenishment strategy; wherein the replenishment strategy classification model is trained based on the third historical sales data of the third time period, the fourth historical sales data of the fourth time period, the preset first replenishment strategy and the preset second replenishment strategy, the third historical sales data of the third time period is used to construct sample feature data of the replenishment strategy classification model, the fourth historical sales data of the fourth time period is used to construct sample label data of the replenishment strategy classification model, the fourth time period is later than the third time period, the second time period is later than the first time period, and the first time period is later than the fourth time period or is the same time period as the fourth time period.

[0117] Optionally, the device also includes a model training module, which is used to: simulate the fourth historical sales data using a preset first replenishment strategy and a preset second replenishment strategy, respectively, to obtain a first simulated inventory indicator and a second simulated inventory indicator; and construct sample label data based on the first simulated inventory indicator and the second simulated inventory indicator.

[0118] Optionally, when the preset first replenishment strategy or the preset second replenishment strategy is a risk value replenishment strategy, the sample feature data further includes attribute feature data, and the model training module is further used to determine the attribute feature data according to the following process:

[0119] Based on the third historical sales data, determine the second quantile value and the average value of the third historical sales data;

[0120] The attribute feature data is calculated according to the second quantile value and the average value of the third historical sales data.

[0121] Optionally, the first simulated inventory index includes a first spot rate and a first inventory turnover days; the second simulated inventory index includes a second spot rate and a second inventory turnover days;

[0122] The model training module is also used to: compare the first spot rate and the second spot rate, the first inventory turnover days and the second inventory turnover days to obtain a first result; and construct the sample label data based on the first result.

[0123] Optionally, the first simulated inventory indicator includes a first target inventory; the second simulated inventory indicator includes a second target inventory;

[0124] The model training module is also used to: determine a first real sales volume corresponding to the first target inventory, and a second real sales volume corresponding to the second target inventory; respectively determine a first difference between the first target inventory and the first real sales volume, and a second difference between the second target inventory and the second real sales volume; and construct the sample label data based on the first difference and the second difference.

[0125] Optionally, the model training module is also used to: respectively determine the positive and negative values ​​of the first difference and the second difference; wherein, if the first target inventory is greater than or equal to the first actual sales volume, the first difference is a positive error, otherwise the first difference is a negative error, and if the second target inventory is greater than or equal to the second actual sales volume, the second difference is a positive error, otherwise the second difference is a negative error; construct the sample label data according to the positive and negative values ​​of the first difference and the second difference.

[0126] The replenishment method of the embodiment of the present invention determines a target replenishment strategy adapted for the target item in the second time period through a pre-trained replenishment strategy classification model and target feature data constructed based on the first historical sales data of the target item in the first time period, and a technical means for determining the replenishment quantity of the target item in the second time period based on the target replenishment strategy. The method can automatically determine a replenishment strategy adapted for the target item, dynamically recommend the most suitable replenishment strategy in a targeted manner, improve replenishment effects, and reduce inventory costs.

[0127] The above device can execute the replenishment method provided by the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, please refer to the method provided by the embodiment of the present invention.

[0128] Figure 5 An exemplary system architecture 500 is shown to which a replenishment method or a replenishment device according to an embodiment of the present invention can be applied.

[0129] like Figure 5 As shown, system architecture 500 may include terminal devices 501, 502, 503, a network 504 and a server 505. Network 504 is used to provide a medium for communication links between terminal devices 501, 502, 503 and server 505. Network 504 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0130] Users can use terminal devices 501, 502, 503 to interact with server 505 through network 504 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 501, 502, 503, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0131] The terminal devices 501 , 502 , and 503 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0132] The server 505 may be a server that provides various services, such as a backend management server that provides support for shopping websites browsed by users using the terminal devices 501, 502, and 503. The backend management server may analyze and process the received data such as product information query requests, and feed back the processing results (such as target push information and product information) to the terminal device.

[0133] It should be noted that the replenishment method provided in the embodiment of the present invention is generally executed by the server 505 , and accordingly, the replenishment device is generally disposed in the server 505 .

[0134] It should be understood that Figure 5 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0135] Reference below Figure 6 , which shows a schematic diagram of the structure of a computer system 600 of a terminal device suitable for implementing an embodiment of the present invention. Figure 6 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0136] like Figure 6 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage part 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the system 600 are also stored. The CPU 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0137] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage section 608 as needed.

[0138] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the central processing unit (CPU) 601, the above-mentioned functions defined in the system of the present invention are executed.

[0139] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0140] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0141] The modules involved in the embodiments of the present invention may be implemented in software or hardware. The modules described may also be set in a processor. For example, they may be described as: a processor includes a sending module, an acquisition module, a determination module, and a first processing module. The names of these modules do not, in some cases, constitute a limitation on the unit itself. For example, the sending module may also be described as a "module for sending a picture acquisition request to the connected server."

[0142] As another aspect, the present invention further provides a computer-readable medium, which may be included in the device described in the above embodiment; or may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device includes:

[0143] Obtain the first historical sales data of the target item in the first time period;

[0144] constructing target feature data based on the first historical sales data;

[0145] Using the pre-trained replenishment strategy classification model and the target feature data, determine the target replenishment strategy adapted for the target item in the second time period, and replenish based on the target replenishment strategy; wherein the replenishment strategy classification model is trained based on the third historical sales data of the third time period, the fourth historical sales data of the fourth time period, the preset first replenishment strategy and the preset second replenishment strategy, the third historical sales data of the third time period is used to construct sample feature data of the replenishment strategy classification model, the fourth historical sales data of the fourth time period is used to construct sample label data of the replenishment strategy classification model, the fourth time period is later than the third time period, the second time period is later than the first time period, and the first time period is later than the fourth time period or is the same time period as the fourth time period.

[0146] The technical solution of the embodiment of the present invention determines a target replenishment strategy adapted for the target item in the second time period through a pre-trained replenishment strategy classification model and target feature data constructed based on the first historical sales data of the target item in the first time period, and a technical means for determining the replenishment quantity of the target item in the second time period based on the target replenishment strategy. The replenishment strategy adapted for the target item can be automatically determined, and the most suitable replenishment strategy can be dynamically recommended in a targeted manner, thereby improving the replenishment effect and reducing inventory costs.

[0147] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A replenishment method, characterized in that: include: Obtain the first historical sales data of the target item in the first time period; Based on the first historical sales data, construct target feature data, wherein the target feature data includes statistical feature data and / or time series feature data; Using a pre-trained replenishment strategy classification model and the target feature data, determine a target replenishment strategy adapted for the target item in the second time period, and perform replenishment based on the target replenishment strategy, wherein the replenishment strategy classification model is trained based on the third historical sales data of the third time period, the fourth historical sales data of the fourth time period, a preset first replenishment strategy, and a preset second replenishment strategy, the third historical sales data of the third time period is used to construct sample feature data of the replenishment strategy classification model, the fourth historical sales data of the fourth time period is used to construct sample label data of the replenishment strategy classification model, the fourth time period is later than the third time period, the second time period is later than the first time period, and the first time period is later than the fourth time period or is the same time period as the fourth time period.

2. The method according to claim 1, characterized in that: The method further includes constructing sample label data of the replenishment strategy classification model according to the following process: The fourth historical sales data is simulated using a preset first replenishment strategy and a preset second replenishment strategy respectively to obtain a first simulated inventory index and a second simulated inventory index; and sample label data is constructed based on the first simulated inventory index and the second simulated inventory index.

3. The method according to claim 2, characterized in that The sample characteristic data includes statistical characteristic data and / or time series characteristic data.

4. The method according to claim 3, characterized in that The statistical feature data includes one or more of the following: historical sales mean, first quantile value, median, standard deviation and maximum value.

5. The method according to claim 3, characterized in that: The time series characteristic data includes one or more of the following: autocorrelation coefficient, partial correlation coefficient, first-order difference, multi-order difference, coefficient of variation and average value of the number of days between non-zero sales.

6. The method according to claim 3, characterized in that When the preset first replenishment strategy or the preset second replenishment strategy is a risk value replenishment strategy, the sample characteristic data further includes attribute characteristic data, wherein the attribute characteristic data is determined according to the following process: Based on the third historical sales data, determine the second quantile value and the average value of the third historical sales data; The attribute feature data is calculated according to the second quantile value and the average value of the third historical sales data.

7. The method according to claim 2, characterized in that The first simulation inventory index includes a first spot rate and a first inventory turnover days; The second simulation inventory index includes a second spot rate and a second inventory turnover days; Based on the first simulation inventory index and the second simulation inventory index, constructing sample label data includes: Comparing the first spot rate and the second spot rate, the first inventory turnover days and the second inventory turnover days to obtain a first result; Based on the first result, the sample label data is constructed.

8. The method according to claim 2, characterized in that: The first simulation inventory indicator includes a first target inventory; The second simulation inventory indicator includes a second target inventory; Based on the first simulation inventory index and the second simulation inventory index, constructing sample label data includes: Determining a first real sales volume corresponding to the first target inventory and a second real sales volume corresponding to the second target inventory; respectively determining a first difference between the first target inventory and the first actual sales volume, and a second difference between the second target inventory and the second actual sales volume; The sample label data is constructed according to the first difference and the second difference.

9. The method according to claim 8, characterized in that Constructing the sample label data according to the first difference and the second difference includes: Determine the positive and negative values ​​of the first difference and the second difference respectively; wherein, if the first target inventory is greater than or equal to the first actual sales volume, the first difference is a positive error, otherwise the first difference is a negative error, and if the second target inventory is greater than or equal to the second actual sales volume, the second difference is a positive error, otherwise the second difference is a negative error; The sample label data is constructed according to the positive and negative values ​​of the first difference and the second difference.

10. A replenishment device, characterized in that: include: A data acquisition module, used to acquire first historical sales data of a target item in a first time period; A feature construction module, configured to construct target feature data based on the first historical sales data, wherein the target feature data includes statistical feature data and / or time series feature data; a replenishment determination module, configured to determine a target replenishment strategy adapted for the target item in the second time period by using a pre-trained replenishment strategy classification model and the target feature data, and to perform replenishment based on the target replenishment strategy; wherein the replenishment strategy classification model is trained based on the third historical sales data of the third time period, the fourth historical sales data of the fourth time period, the preset first replenishment strategy, and the preset second replenishment strategy; the third historical sales data of the third time period is used to construct sample feature data of the replenishment strategy classification model; the fourth historical sales data of the fourth time period is used to construct sample label data of the replenishment strategy classification model; the fourth time period is later than the third time period, the second time period is later than the first time period, and the first time period is later than the fourth time period or is the same time period as the fourth time period.

11. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 9.

12. A computer readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images