A cross-border e-commerce sales forecasting method, device, equipment and storage medium

By obtaining and processing the sales data of cross-border e-commerce, calculating the coefficient of variation and quantiles, eliminating outliers, and combining inventory and trends to predict, the prediction instability of large sales fluctuations and out of stocks in cross-border e-commerce is solved. The problem is to achieve high accuracy and low complexity prediction effects.

CN114548599BActive Publication Date: 2024-11-26NANJING XIYIN ECOMMERCE CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202210220416.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-03-08
Publication Date
2024-11-26
Estimated Expiration
2042-03-08

AI Technical Summary

Technical Problem

The existing technology cannot effectively adapt to scenarios such as large sales fluctuations, wide distribution of sales data, and out of stocks in cross-border e-commerce, resulting in unstable prediction effects.

Method used

By obtaining the basic information of the goods to be replenished and the sales data of the preset days, calculate the coefficient of variation, the first quantile and the second quantile, remove outliers, calculate sales statistical characteristics, and combine inventory and sales trends to Forecasting, avoids gradient descent optimization and is suitable for head and long tail commodities.

Benefits of technology

It realizes stable prediction under low-quality sales data, improves prediction accuracy, reduces algorithm complexity, enhances real-timeness, and avoids delays in stocking process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114548599B_ABST
    Figure CN114548599B_ABST
Patent Text Reader

Abstract

The present application discloses a cross-border e-commerce sales forecasting method, device, equipment and storage medium. The present application first obtains the basic information of the goods to be replenished and the sales data of the preset days, takes into account the sales trend, and also combines it with the sales statistical characteristics and the current inventory to calculate the predicted sales. The present application has low requirements for the quality of sales data, and is suitable for scenarios with a wide distribution range of sales of head goods and long-tail goods. Compared with the existing machine learning methods, there is no need to use the gradient descent method to optimize specific indicators, and there will be no phenomenon of underestimation or overestimation of the forecast. Therefore, the forecast effect is relatively stable and has a certain degree of explainability. In addition, the high prediction accuracy, low algorithm complexity and strong real-time performance can avoid the impact of sales forecast delays on the subsequent stocking process, thereby solving the technical problems that the existing technology cannot adapt to scenarios such as large sales fluctuations, a wide distribution range of sales data and out-of-stock goods, and the forecast effect is unstable.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of commodity sales management, and in particular to a cross-border e-commerce sales forecasting method, device, equipment and storage medium. Background Art

[0002] The cross-border e-commerce industry is developing rapidly, and sales forecasting plays a vital role in the daily operations of merchants. A large sales forecast error will lead to unreasonable calculation of reorder points and replenishment quantities, resulting in too small or too large replenishment quantities. If the replenishment quantity is too small, the goods will be out of stock, resulting in profit losses and reduced service levels due to out-of-stock, while too large a replenishment quantity will increase the turnover days of goods and inventory costs. In addition, unsalable goods will also lead to cost losses.

[0003] Among the existing e-commerce sales forecasting methods, the most common ones are that staff manually calculate the corresponding predicted sales based on the historical average sales of the goods and business experience. There are also time series forecasting methods such as ARIMA, ETS, Prophet, and machine learning forecasting methods such as decision trees and neural networks.

[0004] The cross-border e-commerce industry is currently growing rapidly, with a huge variety and quantity of goods on sale. In addition, due to the "small order, fast return" business model of cross-border e-commerce, there are also characteristics such as fast new product launch and short product life cycle. The method of manually calculating and predicting sales based on business experience is difficult to cope with the situation where the sales trends of different commodities are very different and the sales values ​​are widely distributed. In addition, manual calculation is inefficient and prone to errors, which leads to poor timeliness of stocking. The methods based on time series prediction and machine learning prediction have relatively high requirements for the quality of commodity sales data. Among them, time series prediction usually requires that the sales data is stable and the data length cannot be too short. However, for cross-border e-commerce, the actual sales data may fluctuate greatly, and the short life cycle of some commodities leads to less sales data. For machine learning, it usually uses the gradient descent method to minimize the MAE or MAPE indicators, but the actual sales data has a large distribution range. A large number of long-tail commodities may only have a few or dozens of daily sales, and the head commodities may have a daily sales of hundreds or thousands of pieces. Minimizing MAE and MAPE will cause the prediction model to be biased towards long-tail commodities, and the sales forecast of head commodities will be low. In addition, machine learning is not stable enough, has poor interpretability, and requires a lot of computing resources. Furthermore, existing forecasting methods do not take into account the situation where goods are out of stock. When goods are out of stock, short-term sales may be zero, but sales may be very high in the future. Summary of the invention

[0005] The present application provides a cross-border e-commerce sales forecasting method, device, equipment and storage medium, which are used to solve the technical problems that the existing technology cannot adapt to scenarios such as large sales fluctuations, wide distribution range of sales data and out-of-stock products, and the forecasting effect is unstable.

[0006] The first aspect of the present application provides a cross-border e-commerce sales forecasting method, comprising:

[0007] Obtaining basic information of the product to be replenished and sales data of preset days, wherein the basic information includes the lead time and current inventory of the product to be replenished;

[0008] Calculate the coefficient of variation of the sales data;

[0009] Calculate the first quantile and the second quantile of the sales data;

[0010] Eliminate abnormal values ​​in the sales data according to the first quantile and the second quantile to obtain new sales data;

[0011] Calculate the sales volume statistical characteristics of the to-be-replenished product according to the new sales volume data;

[0012] The predicted sales volume is calculated based on the sales volume statistical characteristics, the coefficient of variation and the basic information.

[0013] This application first obtains the basic information of the goods to be replenished and the sales data of the preset days, takes into account the sales trend of the goods to be replenished, and combines it with the sales statistical characteristics and the current inventory to calculate the predicted sales. This application has low requirements for the quality of sales data, and is suitable for scenarios with a wide distribution range of sales of head goods and long-tail goods. Compared with existing machine learning methods, there is no need to use the gradient descent method to optimize specific indicators, and there will be no phenomenon of underestimation or overestimation of predictions. Therefore, the prediction effect is relatively stable and has a certain degree of explainability. In addition, the high prediction accuracy, low algorithm complexity and strong real-time performance can avoid the impact of sales forecast delays on subsequent stocking processes, thereby solving the technical problems that the existing technology cannot adapt to scenarios such as large sales fluctuations, a wide distribution range of sales data and out-of-stock goods, and the prediction effect is unstable.

[0014] Optionally, the first quantile is the 5% quantile, and the second quantile is the 95% quantile.

[0015] Optionally, calculating the first quantile and the second quantile of the sales data includes:

[0016] Sorting the sales data from small to large to obtain sorted sales data;

[0017] Calculate the first position and the first value of the 5% quantile of the sales data according to the sorted sales data;

[0018] The second position and the second value of the 95% quantile of the sales data are calculated based on the sorted sales data.

[0019] Optionally, the step of eliminating abnormal values ​​in the sales data according to the first quantile and the second quantile to obtain new sales data includes:

[0020] According to the 5% quantile and the 95% quantile, outliers in the sales data are eliminated based on a preset elimination formula to obtain new sales data;

[0021] The preset elimination formula is:

[0022] b n =min(max(a n , Q1), Q3);

[0023] Among them, b n is the new sales data, a n is the sales data, Q1 is the first quantile, Q3 is the second quantile, n is the number of days, min means to remove the smaller value, and max means to take the larger value.

[0024] Optionally, calculating the sales volume statistical features of the to-be-replenished commodity according to the new sales volume data includes:

[0025] Calculate the average sales volume of the new sales volume data in the past 7 days;

[0026] Sort the new sales data from large to small, and calculate the 5-day average sales and 10-day average sales of the products with higher sales.

[0027] Optionally, calculating the coefficient of variation of the sales data includes:

[0028] Calculate the mean and standard deviation of the sales data;

[0029] The coefficient of variation of the sales data is calculated based on the mean and standard deviation.

[0030] Optionally, the calculating the predicted sales volume according to the sales volume statistical characteristics, the coefficient of variation and the basic information includes:

[0031] Dividing a plurality of sales forecast conditions according to the coefficient of variation and the current inventory;

[0032] Based on the sales forecasting conditions, the forecasted sales volume is calculated according to the sales statistical characteristics and the lead time.

[0033] A second aspect of the present application provides a cross-border e-commerce sales forecasting device, comprising:

[0034] An acquisition unit, used to acquire basic information of the commodity to be replenished and sales data of preset days, wherein the basic information includes the lead time and current inventory of the commodity to be replenished;

[0035] A first calculation unit, used to calculate the coefficient of variation of the sales data;

[0036] A second calculation unit, used to calculate the first quantile and the second quantile of the sales data;

[0037] A removal unit, configured to remove abnormal values ​​in the sales data according to the first quantile and the second quantile to obtain new sales data;

[0038] A third calculation unit, configured to calculate the sales volume statistical characteristics of the to-be-replenished commodity according to the new sales volume data;

[0039] The fourth calculation unit is used to calculate the predicted sales volume according to the sales volume statistical characteristics, the coefficient of variation and the basic information.

[0040] A third aspect of the present application provides an electronic device, comprising a processor and a memory storing a computer program, wherein when the processor executes the computer program, the steps of the cross-border e-commerce sales forecasting method as described in the first aspect are implemented.

[0041] The fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the cross-border e-commerce sales forecasting method as described in the first aspect are implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A flow chart of a cross-border e-commerce sales forecasting method provided in an embodiment of the present application;

[0043] Figure 2 A schematic diagram of the sales data change trend of a certain cross-border e-commerce product in the past 15 days provided in the embodiment of the present application;

[0044] Figure 3 A schematic diagram of the structure of a cross-border e-commerce sales forecasting device provided in an embodiment of the present application;

[0045] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] The embodiments of the present application provide a cross-border e-commerce sales forecasting method, device, equipment and storage medium, which are used to solve the technical problems that the existing technology cannot adapt to scenarios such as large sales fluctuations, wide distribution range of sales data and out-of-stock products, and the forecasting effect is unstable.

[0047] In order to make the purpose, features, and advantages of the invention of this application more obvious and easy to understand, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the embodiments described below are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0048] To facilitate understanding of the technical solution of this application, the following is an explanation of the professional terms involved in the technical solution of this application:

[0049] ARIMA: AutoRegressive Integrated Moving Average, a classic time series prediction method.

[0050] ETS: Exponential Smoothing, a commonly used local statistical algorithm for time series forecasting.

[0051] Prophet: A time series prediction algorithm based on trend decomposition proposed by Facebook.

[0052] Lead_time: The replenishment lead time for goods, which is the time interval from when the order is placed to when the goods arrive at the warehouse.

[0053] MAE: Mean Absolute Error, a common prediction evaluation indicator. Generally, absolute error refers to the absolute value of the difference between the actual value and the predicted value. The mean absolute error is the average of the absolute values ​​of the difference.

[0054] MAPE: Mean Absolute Percentage Error, a common prediction error indicator. Generally, the absolute percentage error refers to the ratio of the absolute value of the difference between the actual value and the predicted value to the actual value. The mean absolute percentage error is the average of this ratio.

[0055] See also Figure 1 , the embodiment of the present application provides a cross-border e-commerce sales forecasting method, including:

[0056] S101. Obtain basic information of the product to be replenished and sales data of preset days. The basic information includes the lead time and current inventory of the product to be replenished.

[0057] You can obtain the lead time lead_time and current inventory usable_stk of the goods to be replenished from the data warehouse, and also obtain the sales data of the recent preset days of the goods to be replenished a n , which can be 15 days, 20 days or 30 days. This embodiment uses 15 days of sales data. At the same time, it is also necessary to sort the sales data in reverse order according to time, such as in days, to obtain the preliminary sorted sales data a n , n = 1, 2, 3...15, n represents n days from the current date.

[0058] S102. Calculate the coefficient of variation of sales data.

[0059] It should be noted that the coefficient of variation indicates the relative volatility of sales data. The larger the coefficient of variation, the greater the sales fluctuation, and the smaller the coefficient of variation, the more stable the sales. Therefore, calculating the coefficient of variation of sales data helps to understand the sales trend of the goods to be replenished, in preparation for the subsequent calculation and forecast of sales.

[0060] S103. Calculate the first quantile and the second quantile of the sales data.

[0061] S104. Eliminate abnormal values ​​in the sales data according to the first quantile and the second quantile to obtain new sales data.

[0062] After the sales data is preliminarily sorted out in S101, it is also necessary to eliminate outliers in the sales data through the first quantile and the second quantile to ensure that no outliers appear in the sales data, so that the accuracy and stability of the predicted sales calculated subsequently are high.

[0063] S105. Calculate sales statistics of the to-be-replenished merchandise based on the new sales data.

[0064] S106. Calculate the predicted sales volume based on the sales volume statistical characteristics, coefficient of variation and basic information.

[0065] It should be noted that, in this embodiment, the first quantile may be the 5% quantile, and the second quantile may be the 95% quantile.

[0066] Further, S103 includes:

[0067] S301. Sort the sales data from small to large to obtain sorted sales data.

[0068] In order to improve the accuracy of the calculation, the sales data needs to be sorted from small to large.

[0069] S302. Calculate the first position and the first value of the 5% quantile of the sales data based on the sorted sales data.

[0070] It should be noted that the first position p1 is:

[0071] p1=1+(|{d n}|-1)×0.05;

[0072] The first value is Q1:

[0073]

[0074] Among them, |{d n}| is the data length, symbol Indicates rounding down.

[0075] S303. Calculate the second position and second value of the 95% quantile of the sales data based on the sorted sales data.

[0076] It should be noted that the second position p3 is:

[0077] p3=1+(|{d n}|-1)×0.95;

[0078] The second value is Q3:

[0079]

[0080] Among them, |{d n}| is the data length, symbol Indicates rounding down.

[0081] Further, S104 is specifically:

[0082] According to the 5% quantile and the 95% quantile, the outliers in the sales data are eliminated based on the preset elimination formula to obtain new sales data;

[0083] The preset elimination formula is:

[0084] b n =min(max(a n , Q1), Q3);

[0085] Among them, b n is the new sales data, a n is the sales data, Q1 is the first quantile, Q3 is the second quantile, n is the number of days, for example: n = 1, 2...15, min means to remove the smaller value, and max means to take the larger value.

[0086] Further, S105 includes:

[0087] S501. Calculate the average sales volume of the new sales volume data for the past 7 days.

[0088] This embodiment adopts the 7-day average sales volume. Of course, the technology in this field can also adopt the 5-day or 10-day average sales volume. The 7-day range is generally more reasonable and has the best effect in actual tests. Too long a time cannot reflect the current trend, and too short a time will cause large fluctuations.

[0089] The 7-day sales average is expressed as:

[0090]

[0091] Among them, b n This is the new sales data after removing outliers.

[0092] S502. Sort the new sales data from large to small, and calculate the 5-day average sales and 10-day average sales of the higher sales.

[0093] {b n} Sort by value from large to small to get {c n}.

[0094] The 5-day sales average is expressed as:

[0095]

[0096] The 10-day sales average is expressed as:

[0097]

[0098] Further, S102 includes:

[0099] S201. Calculate the mean and standard deviation of sales data.

[0100] It should be noted that the average value is:

[0101] The standard deviation is:

[0102] S202. Calculate the coefficient of variation of the sales data based on the mean value and standard deviation.

[0103] The coefficient of variation CV is expressed as: CV = σ / μ.

[0104] Further, S106 includes:

[0105] S601. Divide multiple sales forecast conditions according to the coefficient of variation and the current inventory.

[0106] It should be noted that this embodiment includes three sales forecasting conditions, which can be expressed as follows:

[0107] The first sales forecast condition: pred_sales = lead_time × sales_avg_7d;

[0108] The second sales forecast condition: pred_sales = lead_time × sales_avg_top_5d;

[0109] The third sales forecast condition: pred_sales = lead_time × sales_avg_top_10d.

[0110] S602: Based on the sales forecasting conditions, the sales forecast is calculated according to the sales statistical characteristics and the lead time.

[0111] It should be noted that if the coefficient of variation satisfies CV<0.4, the predicted sales volume pred_sales of the product to be replenished during the lead time is calculated according to the first sales volume forecasting condition.

[0112] If the coefficient of variation satisfies 0.4≤CV<0.8, and the current inventory satisfies usable_stk<max(20, 7×sales_avg_7d), where max represents a larger value, the predicted sales pred_sales of the product to be replenished during the lead time is calculated according to the second sales forecast condition.

[0113] If the coefficient of variation satisfies 0.4≤CV<0.8, and the current inventory satisfies usable_stk≥max(20, 7×sales_avg_7d), where max represents a larger value, the predicted sales pred_sales of the to-be-replenished goods during the lead time is calculated according to the first sales forecast condition.

[0114] If the coefficient of variation satisfies CV≥0.8 and the current inventory satisfies usable_stk<max(20, 7×sales_avg_7d), where max represents a larger value, the predicted sales pred_sales of the product to be replenished during the lead time is calculated according to the second sales forecast condition.

[0115] If the coefficient of variation satisfies CV≥0.8 and the current inventory satisfies usable_stk≥max(20, 7×sales_avg_7d), where max represents a larger value, the predicted sales volume pred_sales of the product to be replenished during the lead time is calculated according to the third sales forecast condition.

[0116] This embodiment first obtains the basic information of the goods to be replenished and the sales data of the preset days, takes into account the sales trend of the goods to be replenished, and also combines it with the sales statistical characteristics and the current inventory to calculate the predicted sales. This application has low requirements for the quality of sales data, and is suitable for scenarios with a wide distribution range of sales of head goods and long-tail goods. Compared with existing machine learning methods, there is no need to use the gradient descent method to optimize specific indicators, and there will be no phenomenon of underestimation or overestimation of predictions. Therefore, the prediction effect is relatively stable and has a certain degree of interpretability. In addition, the high prediction accuracy, low algorithm complexity and strong real-time performance can avoid the impact of sales forecast delays on subsequent stocking processes, thereby solving the technical problems that the existing technology cannot adapt to scenarios such as large sales fluctuations, a wide distribution range of sales data and out-of-stock goods, and the prediction effect is unstable.

[0117] The above is a detailed description of an embodiment of a cross-border e-commerce sales forecasting method provided by the present application. The following is a detailed description of an embodiment of a cross-border e-commerce sales forecasting device provided by the present application. The cross-border e-commerce sales forecasting device described below and the cross-border e-commerce sales forecasting method described above can be referenced to each other.

[0118] See also Figure 3 The present application embodiment provides a cross-border e-commerce sales forecasting device, including:

[0119] An acquisition unit 410 is used to acquire basic information of the commodity to be replenished and sales data of preset days, wherein the basic information includes the lead time and current inventory of the commodity to be replenished;

[0120] A first calculation unit 420, used to calculate the coefficient of variation of the sales data;

[0121] The second calculation unit 430 is used to calculate the first quantile and the second quantile of the sales data;

[0122] A removal unit 440, configured to remove abnormal values ​​in the sales data according to the first quantile and the second quantile to obtain new sales data;

[0123] The third calculation unit 450 is used to calculate the sales statistical characteristics of the commodity to be replenished according to the new sales data;

[0124] The fourth calculation unit 460 is used to calculate the predicted sales volume according to the sales volume statistical characteristics, the coefficient of variation and the basic information.

[0125] Further, the first quantile is the 5% quantile, and the second quantile is the 95% quantile.

[0126] Further, the second calculation unit 430 includes:

[0127] The sorting subunit is used to sort the sales data from small to large to obtain sorted sales data.

[0128] The first calculation subunit is used to calculate the first position and the first value of the 5% quantile of the sales data according to the sorted sales data.

[0129] The second calculation subunit is used to calculate the second position and second value of the 95% quantile of the sales data according to the sorted sales data.

[0130] Further, the elimination unit 440 is specifically used to eliminate abnormal values ​​in the sales data according to the 5% quantile and the 95% quantile based on a preset elimination formula to obtain new sales data;

[0131] The preset elimination formula is:

[0132] b n =min(max(a n , Q1), Q3);

[0133] Among them, b n is the new sales data, a n is the sales data, Q1 is the first quantile, Q3 is the second quantile, n is the number of days, min means to remove the smaller value, and max means to take the larger value.

[0134] Further, the third calculation unit 450 includes:

[0135] The third calculation subunit is used to calculate the average sales volume of the new sales data in the past 7 days.

[0136] The fourth calculation subunit is used to sort the new sales data from large to small, and calculate the 5-day average sales volume and the 10-day average sales volume of the higher sales volume.

[0137] Further, the first calculation unit 420 includes:

[0138] The fifth calculation subunit is used to calculate the average value and standard deviation of the sales data.

[0139] The sixth calculation subunit is used to calculate the coefficient of variation of the sales data based on the mean value and the standard deviation.

[0140] Further, the fourth calculation unit 460 includes:

[0141] Divide subunits to divide multiple sales forecast conditions based on coefficient of variation and current inventory;

[0142] The seventh calculation subunit is used to calculate the predicted sales volume based on the sales volume prediction conditions, according to the sales volume statistical characteristics and the lead time.

[0143] Figure 4 The following is a schematic diagram of the physical structure of an electronic device. Figure 4 As shown, the present invention also provides an electronic device, which may include: a processor (processor) 310, a communication interface (Communication Interface) 320, a memory (memory) 330 and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other through the communication bus 340. The processor 310 can call the computer program in the memory 330 to execute the steps of a cross-border e-commerce sales prediction method, for example including:

[0144] Get the basic information of the products to be replenished and the sales data of the preset days. The basic information includes the lead time and current inventory of the products to be replenished.

[0145] Calculate the coefficient of variation of sales data;

[0146] Calculate the first and second quantiles of sales data;

[0147] Eliminate outliers in the sales data based on the first and second quantiles to obtain new sales data;

[0148] Calculate the sales statistics of the products to be replenished based on the new sales data;

[0149] Calculate the predicted sales volume based on sales statistics, coefficient of variation and basic information.

[0150] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0151] On the other hand, an embodiment of the present application further provides a computer-readable storage medium, wherein the processor-readable storage medium stores a computer program, and the computer program is used to enable the processor to execute the steps of the methods provided in the above embodiments, for example, including:

[0152] Get the basic information of the products to be replenished and the sales data of the preset days. The basic information includes the lead time and current inventory of the products to be replenished.

[0153] Calculate the coefficient of variation of sales data;

[0154] Calculate the first and second quantiles of sales data;

[0155] Eliminate outliers in the sales data based on the first and second quantiles to obtain new sales data;

[0156] Calculate the sales statistics of the products to be replenished based on the new sales data;

[0157] Calculate the predicted sales volume based on sales statistics, coefficient of variation and basic information.

[0158] The processor-readable storage medium can be any available medium or data storage device that can be accessed by the processor, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor storage (such as ROM, EPROM, EEPROM, non-volatile memory (NANDFLASH), solid-state drive (SSD)), etc.

[0159] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0160] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0161] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A cross-border e-commerce sales forecasting method, characterized in that: include: Obtaining basic information of the product to be replenished and sales data of preset days, wherein the basic information includes the lead time and current inventory of the product to be replenished; Calculate the coefficient of variation of the sales data, including: Calculate the mean and standard deviation of the sales data; Calculate the coefficient of variation of the sales data based on the mean value and standard deviation; Calculating the first quantile and the second quantile of the sales data, where the first quantile is the 5% quantile and the second quantile is the 95% quantile, includes: Sorting the sales data from small to large to obtain sorted sales data; Calculate the first position and the first value of the 5% quantile of the sales data according to the sorted sales data; Calculate the second position and second value of the 95% quantile of the sales data according to the sorted sales data; Eliminating abnormal values ​​in the sales data according to the first quantile and the second quantile to obtain new sales data includes: According to the 5% quantile and the 95% quantile, outliers in the sales data are eliminated based on a preset elimination formula to obtain new sales data; The preset elimination formula is: b n =min(max(a n ,Q1),Q3); Among them, b n is the new sales data, a n is the sales data, Q1 is the first quantile, Q3 is the second quantile, n is the number of days, min means to remove the smaller value, and max means to take the larger value; Calculating the sales statistics of the to-be-replenished product according to the new sales data includes: Calculate the average sales volume of the new sales volume data in the past 7 days; Sort the new sales data from large to small, and calculate the 5-day average sales and 10-day average sales of the products with higher sales; Calculating the predicted sales volume according to the sales volume statistical characteristics, the coefficient of variation and the basic information includes: Dividing a plurality of sales forecast conditions according to the coefficient of variation and the current inventory; Based on the sales forecasting conditions, the forecasted sales volume is calculated according to the sales statistical characteristics and the lead time.

2. A cross-border e-commerce sales forecasting device, characterized in that: include: An acquisition unit, used to acquire basic information of the commodity to be replenished and sales data of preset days, wherein the basic information includes the lead time and current inventory of the commodity to be replenished; A first calculation unit, used to calculate the coefficient of variation of the sales data; A second calculation unit, used to calculate the first quantile and the second quantile of the sales data; A removal unit, configured to remove abnormal values ​​in the sales data according to the first quantile and the second quantile to obtain new sales data; A third calculation unit, configured to calculate the sales volume statistical characteristics of the to-be-replenished commodity according to the new sales volume data; The fourth calculation unit is used to calculate the predicted sales volume according to the sales volume statistical characteristics, the coefficient of variation and the basic information.

3. An electronic device comprising a processor and a memory storing a computer program, characterized in that: When the processor executes the computer program, the steps of the cross-border e-commerce sales prediction method as described in claim 1 are implemented.

4. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the steps of the cross-border e-commerce sales forecasting method as claimed in claim 1 are implemented.

Citation Information

Patent Citations

  • Method and apparatus for determining peak sale and one year sale prediction for hardcover first releases

    EP3418916A1