Cross-border e-commerce replenishment management system based on big data

Abstract

The invention relates to the technical field of e-commerce replenishment, in particular to a cross-border e-commerce replenishment management system based on big data, which comprises a data interface connection module, a sales data analysis and prediction module, an inventory monitoring and replenishment decision module and a sales prediction and adjustment module, the inventory monitoring and replenishment decision-making module judges the replenishment date of the commodity A by sensing the inventory number of the commodity A in the cross-border e-commerce system, predicts the change trend of the daily sales volume of the commodity A in the current time interval through the commodity sales volume prediction unit, and calculates the replenishment number of the commodity A. Accurate inventory management and replenishment decision-making are realized. According to real-time inventory and accurate sales volume prediction, replenishment can be performed at a proper time, and a reasonable replenishment quantity can be determined. The problems of stock-out in the peak season and inventory overstock in the slack season caused by inaccurate experience judgment in a traditional method are avoided, the inventory cost is reduced, and the operation efficiency and customer satisfaction of enterprises are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of e-commerce replenishment, and in particular to a cross-border e-commerce replenishment management system based on big data. Background Art

[0002] Cross-border e-commerce faces consumers all over the world, whose shopping habits and demands vary. They also have high expectations for the availability of goods. If a product is out of stock after a user places an order, it will affect the user's shopping experience. Therefore, timely replenishment of goods is crucial.

[0003] At present, many cross-border e-commerce systems rely on staff to set inventory thresholds based on their own experience. Once the inventory of a product falls below this threshold, the replenishment process will be initiated. However, the peak sales seasons of different products are different. Setting inventory thresholds based solely on staff experience is bound to have many disadvantages. On the one hand, insufficient estimates of the peak sales season may lead to a serious shortage of inventory during the peak season, which not only makes it impossible to meet consumers' shopping needs, but also damages the store's reputation and sales. On the other hand, during the off-season, unreasonable threshold settings may cause inventory backlogs, occupy a large amount of funds and storage space, and increase operating costs. In view of this, we propose a cross-border e-commerce replenishment management system based on big data. Summary of the invention

[0004] The purpose of the present invention is to solve the problems existing in the commodity replenishment link of the current cross-border e-commerce system. When replenishing different commodities, the existing system usually relies on staff to set inventory thresholds based on personal experience. However, the sales peak and off-peak seasons of various commodities are different. In the off-season, the commodity sales cycle is long, resulting in the problem that in the peak season, the staff determines the replenishment time and replenishment quantity based on the off-season sales.

[0005] To achieve the above-mentioned purpose, the present invention provides a cross-border e-commerce replenishment management system based on big data, including a data interface connection module, a sales data analysis and forecasting module, an inventory monitoring and replenishment decision module, and a sales forecast adjustment module;

[0006] The sales data analysis and prediction module includes a time interval division unit and a commodity sales volume prediction unit;

[0007] The time interval division unit is used to set a hot-selling threshold and a time threshold, and record the number of days when the historical daily sales volume of commodity A is greater than the hot-selling threshold of the corresponding year. If the number of days is greater than the time threshold, the hot-selling days are connected in chronological order, and the continuous order time is merged into a time interval, which is defined as a hot-selling time interval. Otherwise, it is defined as a cold-selling time interval.

[0008] The commodity sales forecasting unit uses an exponential smoothing method to forecast the hot-selling time interval of commodity A in the current year and the sales volume of commodity A every day in the current time interval. The exponential smoothing method assigns a fixed weight to each historical year and forecasts according to the corresponding weight;

[0009] The inventory monitoring and replenishment decision module senses the inventory quantity of commodity A in the cross-border e-commerce system, determines the date threshold according to the sales volume of A, replenishes commodity A, and calculates the replenishment quantity of commodity A at different time nodes;

[0010] The sales forecast adjustment module is used to call out the daily sales of product A corresponding to the previous time interval and the daily sales of each time interval in the historical order data, which are respectively defined as sales set A and multiple sales sets B, match and calculate the similarity between sales set A and multiple sales sets B, call out the year corresponding to the sales set B with the maximum similarity, define the year as the first year, and adjust the weights of different years in the exponential smoothing method.

[0011] As a further improvement of the present technical solution, the data interface connection module establishes a connection with the cross-border e-commerce system through a data interface, uses a random function to extract the historical order data of any commodity A in the database, and then uses the year function and sales volume function to call out the year and sales volume of commodity A in the historical order data.

[0012] As a further improvement of the technical solution, the data interface connection module constructs a request according to the data interface specification of the cross-border e-commerce system, the request includes verification information, and then sends the constructed request to the data interface address of the cross-border e-commerce system using the TCP / IP protocol;

[0013] After receiving the request, the data interface of the cross-border e-commerce system parses the request according to the pre-set parsing rules to verify whether the request format is standard. If it is standard, the data interface connection module establishes a connection with the cross-border e-commerce system.

[0014] As a further improvement of the technical solution, the core of the random function is to generate a pseudo-random number in the historical local data. When the historical order data of commodity A is extracted from the database, the random function assigns a random number to each order record, and then selects any order data by the random number;

[0015] Date functions are used to process and operate date and time data. They can identify the date format stored in the database and extract the various parts of the date according to the requirements of the data interface connection module.

[0016] The sales volume function is used to count and calculate the sales quantity of a product. It traverses the order records related to product A in the database, identifies and accumulates the sales quantity of product A in each order, matches product A according to the product number in the order data, and then obtains the corresponding sales quantity field from the order record for accumulation, thereby obtaining the sales volume of product A.

[0017] As a further improvement of the technical solution, the time interval division unit groups the same year information of commodity A in the historical order data into one group, and senses the historical daily sales volume of each group of commodity A and the total sales volume of each group of commodity A in the corresponding year;

[0018] Divide the total sales of product A in each year by the number of days in the corresponding year to get the average daily sales of product A, which is defined as the hot-selling threshold of the product in the corresponding year;

[0019] Traverse the historical daily sales of product A in the corresponding year, record the number of days when the historical daily sales are greater than the hot-selling threshold of the corresponding year, and define it as the hot-selling days of the product.

[0020] As a further improvement of the technical solution, the time interval division unit sets the time threshold using the median setting method: the number of hot-selling days of the product in the previous year is sensed, the hot-selling days of different products are sorted from large to small, and the median is taken as the time threshold. The corresponding calculation formula is as follows:

[0021] The number of days that the product was perceived to be hot-selling in the previous year is: ,in The number of times the daily sales volume in the previous year was greater than the hot-selling threshold, For the The number of days when the sales volume is greater than the hot selling threshold, After sorting from largest to smallest, we get: ,in ;

[0022] like When it is an odd number, the median is the element in the middle of the sorted set, that is ;

[0023] like Even number, median is the average of the two middle elements of the sorted set, that is, .

[0024] As a further improvement of the technical solution, the exponential smoothing method in the commodity sales forecasting unit performs weighted averages on the hot-selling time intervals and the cold-selling time intervals of each year. For time intervals of different years, the closer the year is to the current year, the greater the impact of the corresponding time interval on the prediction of the current time interval, so the higher the weight assigned; the farther the year is from the current year, the smaller the impact of the corresponding time interval on the prediction of the current time interval, and the lower the weight assigned, and the year weight decays exponentially. By using the decay factor to control the decay speed of the weight, the calculation formula for predicting the hot-selling time interval by the exponential smoothing method is as follows:

[0025] Perception for The hot selling time period of product A in the year, for The predicted hot-selling time period of commodity A in the year, is the smoothing coefficient, are the weights corresponding to different years, and ,in The attenuation factor takes a value between 0 and 1. is the current year, is the corresponding year, then predict the current year Hot selling time period The formula is ,in is the number of historical years;

[0026] Compare the current time with the predicted hot-selling time interval of product A in the current year:

[0027] If the current time is If the product A is in the hot-selling time range,

[0028] If the current time is not range, then product A is in the cold-selling time period.

[0029] As a further improvement of the technical solution, the inventory monitoring and replenishment decision module uses a line chart analysis method to analyze the daily sales quantity change trend predicted by the commodity sales forecasting unit within the current time interval;

[0030] Draw a line graph with the date in the current time interval as the horizontal axis and the predicted daily sales quantity as the vertical axis. If the line graph is tilted upward, it is determined that the sales quantity is in an upward trend. If the line graph is tilted downward as a whole, it is determined that the sales quantity is in a downward trend.

[0031] If the trend of change is in an upward trend, the date corresponding to the maximum sales volume is called out, the total sales volume of commodity A between the current date and the maximum sales volume date is calculated, it is set as the replenishment inventory quantity, and the replenishment inventory quantity signal is output at the same time.

[0032] As a further improvement of the technical solution, the sales forecast adjustment module senses the current time node of commodity A, and adjusts the daily sales of commodity A in the corresponding previous time interval, and all daily sales in the previous time interval are sales set A, and the historical order data of commodity A is divided into time intervals through the time interval division unit, and the time interval corresponding to the previous time interval is called out, which is defined as the matching time interval, and the daily sales of the commodity in each matching time interval is the sales set B;

[0033] The similarity between the sales set A and multiple sales sets B is matched and calculated, and the year corresponding to the sales set B with the maximum similarity is retrieved, and the year is defined as the first year. When the commodity sales forecasting unit adopts the exponential smoothing method to predict the sales quantity of commodity A every day in the current time interval, the weight of the previous year at the current time node is replaced with the weight of the first year, and the weight of the previous year is replaced with the weight of the year before that, until all years of commodity A are replaced.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] In this cross-border e-commerce replenishment management system based on big data, the different time nodes of commodity A in the historical order data are divided by the time interval division unit, and then the commodity sales prediction unit uses the exponential smoothing method to predict the time interval corresponding to commodity A in the current year and the daily sales of commodity A in each time interval. At this time, the inventory monitoring and replenishment decision module determines the date of replenishment of commodity A by sensing the inventory quantity of commodity A in the cross-border e-commerce system, and predicts the change trend of daily sales of commodity A in the current time interval through the commodity sales prediction unit, calculates the replenishment quantity of commodity A, and realizes accurate inventory management and replenishment decision. It can replenish at the right time and determine the reasonable replenishment quantity based on real-time inventory and accurate sales forecasts, avoiding the problems of out-of-stock in the peak season and inventory backlog in the off-season caused by inaccurate experience judgment in traditional methods, reducing inventory costs, and improving the company's operating efficiency and customer satisfaction.

[0036] The sales forecast adjustment module calls out the previous time interval corresponding to the current time node, and the time interval division unit divides the time interval in the historical data order, matches and calculates the similarity of product A in the previous time interval and the same interval in the historical data order, and adjusts the weight of the exponential smoothing method in the sales data analysis and forecasting module to predict the corresponding year of product sales through the similarity, further optimizing the accuracy of sales forecasts. Taking into account that sales in the same time period in different years may differ, the forecast weight is adjusted through similarity analysis to make the forecast result more in line with the actual market situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is the overall module principle diagram of the present invention;

[0038] Figure 2 It is a flow chart of the working principle of the present invention.

[0039] The meaning of each number in the figure is:

[0040] 100. Data interface connection module; 200. Sales data analysis and forecasting module; 210. Time interval division unit; 220. Commodity sales forecasting unit; 300. Inventory monitoring and replenishment decision module; 400. Sales forecasting adjustment module. DETAILED DESCRIPTION

[0041] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] like Figure 1-Figure 2 As shown, the cross-border e-commerce replenishment management system based on big data includes a data interface connection module 100, a sales data analysis and prediction module 200, an inventory monitoring and replenishment decision module 300 and a sales forecast adjustment module 400;

[0043] As the existing cross-border e-commerce system involves many participants, such as suppliers, logistics companies, payment platforms, etc., in order to facilitate the connection between upstream and downstream companies, data interfaces will be pre-defined, which will enable the cross-border e-commerce system to be integrated with external companies quickly and efficiently, and realize data sharing and interaction;

[0044] The data interface connection module 100 establishes a connection with the cross-border e-commerce system through a data interface, and uses a random function to extract the historical order data of any commodity A in the database (the order data includes information such as the order number, the commodity number, and the order time), and a year function and a sales volume function to call out the year and sales volume of commodity A in the historical order data;

[0045] The data interface connection module 100 constructs a request according to the data interface specification of the cross-border e-commerce system, the request includes verification information, and then sends the constructed request to the data interface address of the cross-border e-commerce system using the TCP / IP protocol;

[0046] After receiving the request, the data interface of the cross-border e-commerce system parses the request according to the pre-set parsing rules to verify whether the request format is standardized. If it is standardized, the data interface connection module 100 establishes a connection with the cross-border e-commerce system;

[0047] The core of the random function is to generate pseudo-random numbers in historical local data. When it is used to extract the historical order data of commodity A in the database, the random function assigns a random number to each order record, and then selects any order data through the random number;

[0048] The date function is used to process and operate date and time data, and can identify the date format stored in the database and extract various parts of the date, such as year, month, day, etc., according to the requirements of the data interface connection module 100. The date function treats the date data as a specific value or data structure based on specific date and time standards and algorithms;

[0049] The sales volume function is used to count and calculate the sales quantity of a product. It will traverse the order records related to product A in the database, identify and accumulate the sales quantity of product A in each order, match product A according to the product number in the order data, and then obtain the corresponding sales quantity field from the order record for accumulation operation to obtain the sales volume of product A.

[0050] Cross-border e-commerce faces consumers all over the world, whose shopping habits and demands vary. They also have high expectations for the availability of goods. If a product is out of stock after a user places an order, it will affect the user's shopping experience. Therefore, timely replenishment of goods is crucial.

[0051] At present, many cross-border e-commerce systems rely on staff to set inventory thresholds based on their own experience. Once the inventory of a product falls below this threshold, the replenishment process will be initiated. However, the peak sales seasons of different products are different. Setting inventory thresholds based solely on staff experience is bound to have many disadvantages. On the one hand, due to insufficient estimation of the peak sales season, there may be a serious shortage of inventory during the peak season, which not only makes it impossible to meet consumers' shopping needs, but also damages the store's reputation and sales. On the other hand, during the off-season, the threshold setting may be unreasonable, resulting in inventory backlogs, occupying a large amount of funds and storage space, and increasing operating costs.

[0052] The sales data analysis and prediction module 200 includes a time interval division unit 210 and a commodity sales volume prediction unit 220;

[0053] The time interval division unit 210 groups the same year information of commodity A in the historical order data into one group, and senses the historical daily sales volume of each group of commodity A and the total sales volume of each group of commodity A in the corresponding year;

[0054] Divide the total sales of product A in each year by the number of days in the corresponding year to get the average daily sales of product A, which is defined as the hot-selling threshold of the product in the corresponding year. The calculation formula is: ;

[0055] in Indicates Number of days in a year (common year ,leap year ), Indicates Item No. The hot selling threshold of the year;

[0056] Traverse the historical daily sales of product A in the corresponding year, record the number of days when the historical daily sales are greater than the hot-selling threshold of the corresponding year, and define it as the hot-selling days of the product.

[0057] The time interval division unit 210 also uses the median setting method to set the time threshold:

[0058] Perceive the number of hot-selling days of the product in the previous year, sort the hot-selling days of different products from large to small, and take the median as the time threshold. The corresponding calculation formula is as follows:

[0059] The number of days that the product was perceived to be hot-selling in the previous year is: ,in The number of times the daily sales volume in the previous year was greater than the hot-selling threshold, For the The number of days when the sales volume is greater than the hot selling threshold, After sorting from largest to smallest, we get: ,in ;

[0060] like When it is an odd number, the median is the element in the middle of the sorted set, that is ;

[0061] like Even number, median is the average of the two middle elements of the sorted set, that is, ;

[0062] If the number of hot-selling days of the product in the current year is greater than the time threshold, the hot-selling days will be connected in chronological order, and the continuous order times will be merged into a time interval, which is defined as the hot-selling time interval. Then, the dates that are not outside the hot-selling time interval will be connected in chronological order and defined as the cold-selling time interval.

[0063] The present invention fully considers that there are obvious off-season and peak-season differences in the sales process of goods. In the off-season, the market demand is low, and the inventory of goods can be relatively maintained at a low level; in the peak season, the market demand increases significantly, and the required inventory of goods will increase significantly. If the demand for goods in different time periods cannot be accurately predicted in the process of managing the inventory of goods, it will be difficult to determine a reasonable inventory threshold, which will lead to a large number of sales opportunities missed due to insufficient inventory during the peak sales season, resulting in sales losses, and will also affect consumers' shopping experience and damage the reputation of the store; in the off-season, too much inventory backlog may occupy a large amount of funds and storage space, increasing operating costs;

[0064] The commodity sales prediction unit 220 uses the exponential smoothing method to predict the hot-selling time interval of commodity A in the current year, and determines the time interval of commodity A at the current time node, and then uses the exponential smoothing method again to predict the sales volume of commodity A every day in the current time interval;

[0065] The exponential smoothing method performs weighted average of the hot-selling time interval and the cold-selling time interval each year. For time intervals of different years, the closer the year is to the current year, the greater the impact of the corresponding time interval on the prediction of the current time interval, so the higher the weight is assigned; the farther the year is from the current year, the smaller the impact of the corresponding time interval on the prediction of the current time interval, and the lower the weight is assigned. For example, assuming the current year is , the previous year was , the previous year was etc., then The weight of the time interval corresponding to the year> The weight of the time interval corresponding to the year;

[0066] The year weight decays exponentially. By using the decay factor to control the decay speed of the weight, the calculation formula for predicting the hot-selling time interval by the exponential smoothing method is as follows:

[0067] Perception for The hot selling time period of product A in the year, for The predicted hot-selling time period of commodity A in the year, is the smoothing coefficient ( ), are the weights corresponding to different years, and ,in The attenuation factor takes a value between 0 and 1. is the current year, is the corresponding year, then predict the current year Hot selling time period The formula is ,in is the number of historical years;

[0068] Compare the current time with the predicted hot-selling time interval of product A in the current year:

[0069] If the current time is If the product A is in the hot-selling time range,

[0070] If the current time is not range, then product A is in the cold-selling time period.

[0071] The inventory monitoring and replenishment decision module 300 senses the inventory quantity of commodity A in the cross-border e-commerce system, and the commodity sales prediction unit 220 predicts the sales quantity of commodity A corresponding to the current time node;

[0072] If the sales quantity of commodity A on date A is less than the sales quantity on date A, date A is determined to be the date threshold, and a signal that commodity A needs to be replenished is outputted through the connection established between the data interface connection module 100 and the cross-border e-commerce system;

[0073] Using a line graph analysis method to analyze the daily sales quantity change trend predicted by the commodity sales prediction unit 220 within the current time interval;

[0074] Draw a line graph with the date in the current time interval as the horizontal axis and the predicted daily sales quantity as the vertical axis. If the line graph is tilted upward, it is determined that the sales quantity is in an upward trend. If the line graph is tilted downward as a whole, it is determined that the sales quantity is in a downward trend.

[0075] If the trend is rising, call up the date corresponding to the maximum sales volume, calculate the total sales volume of commodity A between the current date and the maximum sales volume date, set it as the replenishment inventory quantity, and output the replenishment inventory quantity signal at the same time. The corresponding calculation formula is as follows:

[0076] Sense current date is , the commodity sales forecasting unit 220 predicts the maximum sales date to be , the commodity sales volume forecasting unit 220 forecasts The forecast sales for each day in the interval are ;

[0077] Replenishment of stock quantity , which is the sum of the predicted sales for each day from the current date to the maximum sales date;

[0078] If it is in a downward trend, calculate the decline rate during the downward trend. The decline rate is the ratio of the average predicted sales before the downward trend starts to the average predicted sales after the start of the downward trend at the same time node. The calculation formula is: ;

[0079] Then, the replenishment inventory quantity is reduced in proportion to the decline, and the replenishment inventory quantity calculated under the perceived upward trend is , then the adjusted replenishment inventory quantity .

[0080] The present invention takes into account that during the sales process of commodities, the exponential smoothing method in the commodity sales forecasting unit 220 assigns a fixed weight to each historical year, but the market situation is dynamically changing, and the sales situation in some years may be more similar to the current market environment. The fixed weight cannot reflect this difference, which affects the accuracy of the forecast result;

[0081] The sales forecast adjustment module 400 senses the current time node of commodity A, and adjusts the daily sales of commodity A in the previous time interval, defines all daily sales in the previous time interval as sales set A, and divides the historical order data of commodity A into time intervals through the time interval division unit 210, and calls out the time interval corresponding to the previous time interval, which is defined as the matching time interval, and the daily sales of the commodity in each matching time interval is defined as sales set B;

[0082] The similarity between the sales volume set A and the multiple sales volume sets B is matched and calculated, and the year corresponding to the sales volume set B with the maximum similarity is retrieved, and the year is defined as the first year. When the commodity sales volume forecasting unit 220 uses the exponential smoothing method to forecast the sales volume of commodity A every day in the current time interval, the weight of the previous year at the current time node is replaced with the weight of the first year, and the weight of the previous year is replaced with the weight of the year before that, until all the years of commodity A are replaced. Since the first year, which is the historical year most similar to the current sales situation, has a greater impact on the current sales volume forecast, it should be given a higher weight.

[0083] The calculation formula for the similarity between sales set A and sales set B is as follows:

[0084] Sales Collection , sales collection ,in Respectively represent the number of elements in sales set A and sales set B;

[0085] The Euclidean distance between sales set A and sales set B is: , where the smaller the Euclidean distance, the more similar the two sets are.

[0086] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A cross-border e-commerce replenishment management system based on big data, characterized by: The system comprises a data interface connection module (100) for retrieving historical data, a sales data analysis and forecasting module (200), an inventory monitoring and replenishment decision module (300) and a sales forecast adjustment module (400), wherein: The sales data analysis and prediction module (200) comprises a time interval division unit (210) and a commodity sales volume prediction unit (220); the time interval division unit (210) is used to set a hot-selling threshold and a time threshold, and record the number of days when the historical daily sales volume of commodity A is greater than the hot-selling threshold of the corresponding year. If the number of days is greater than the time threshold, the hot-selling days are connected in chronological order, and the continuous order time is merged into a time interval, which is defined as a hot-selling time interval, otherwise it is defined as a cold-selling time interval; the commodity sales volume prediction unit (220) uses an exponential smoothing method to predict the hot-selling time interval of commodity A in the current year and the sales volume of commodity A every day in the current time interval. The exponential smoothing method assigns a fixed weight to each historical year, and makes a prediction based on the corresponding weight; The inventory monitoring and replenishment decision module (300) senses the inventory quantity of commodity A in the cross-border e-commerce system, determines the date threshold according to the sales volume of A, replenishes commodity A, and calculates the replenishment quantity of commodity A at different time nodes; The sales forecast adjustment module (400) is used to retrieve the daily sales of commodity A corresponding to the previous time interval and the daily sales of each time interval in the historical order data, which are respectively defined as a sales set A and multiple sales sets B, match and calculate the similarity between the sales set A and the multiple sales sets B, retrieve the year corresponding to the sales set B with the maximum similarity, define the year as the first year, and adjust the weights of different years in the exponential smoothing method.

2. The cross-border e-commerce replenishment management system based on big data according to claim 1 is characterized by: The data interface connection module (100) establishes a connection with the cross-border e-commerce system via a data interface, uses a random function to extract historical order data of any commodity A in the database, and then uses a year function and a sales volume function to retrieve the year and sales volume of commodity A in the historical order data.

3. The cross-border e-commerce replenishment management system based on big data according to claim 2 is characterized by: The data interface connection module (100) constructs a request according to the data interface specification of the cross-border e-commerce system, the request including verification information, and then sends the constructed request to the data interface address of the cross-border e-commerce system using the TCP / IP protocol; After receiving the request, the data interface of the cross-border e-commerce system parses the request according to pre-set parsing rules to verify whether the request format is standard. If it is standard, the data interface connection module (100) establishes a connection with the cross-border e-commerce system.

4. The cross-border e-commerce replenishment management system based on big data according to claim 3 is characterized by: The core of the random function is to generate pseudo-random numbers in historical local data. When it is used to extract the historical order data of commodity A in the database, the random function assigns a random number to each order record, and then selects any order data through the random number; The date function is used to process and operate date and time data, and is capable of identifying the date format stored in the database and extracting various parts of the date according to the requirements of the data interface connection module (100); The sales volume function is used to count and calculate the sales quantity of a product. It traverses the order records related to product A in the database, identifies and accumulates the sales quantity of product A in each order, matches product A according to the product number in the order data, and then obtains the corresponding sales quantity field from the order record for accumulation, thereby obtaining the sales volume of product A.

5. The cross-border e-commerce replenishment management system based on big data according to claim 4 is characterized by: The time interval division unit (210) groups the same year information of commodity A in the historical order data into one group, and senses the historical daily sales volume of each group of commodity A and the total sales volume of each group of commodity A in the corresponding year; Divide the total sales of product A in each year by the number of days in the corresponding year to get the average daily sales of product A, which is defined as the hot-selling threshold of the product in the corresponding year; Traverse the historical daily sales of product A in the corresponding year, record the number of days when the historical daily sales are greater than the hot-selling threshold of the corresponding year, and define it as the hot-selling days of the product.

6. The cross-border e-commerce replenishment management system based on big data according to claim 5 is characterized by: The time interval division unit (210) sets the time threshold using the median setting method: the number of hot-selling days of the product in the previous year is sensed, the hot-selling days of different products are sorted from large to small, and the median is taken as the time threshold. The corresponding calculation formula is as follows: The perceived number of days that products were hot sellers in the previous year is: ,in The number of times the daily sales volume in the previous year was greater than the hot-selling threshold, For the The number of days when the sales volume is greater than the hot selling threshold, After sorting from largest to smallest, we get: ,in ; like When it is an odd number, the median is the element in the middle of the sorted set, that is ; like Even number, median is the average of the two middle elements of the sorted set, that is, .

7. The cross-border e-commerce replenishment management system based on big data according to claim 1 is characterized by: The exponential smoothing method in the commodity sales prediction unit (220) performs weighted average on the hot-selling time interval and the cold-selling time interval of each year. For time intervals of different years, the closer the year is to the current year, the greater the impact of the corresponding time interval on the prediction of the current time interval, so the weight assigned is higher; the farther the year is from the current year, the smaller the impact of the corresponding time interval on the prediction of the current time interval, and the lower the weight assigned. The year weight decays exponentially. By using the decay factor to control the decay speed of the weight, the calculation formula for predicting the hot-selling time interval by the exponential smoothing method is as follows: Perception for The hot selling time period of product A in the year, for The predicted hot-selling time period of commodity A in the year, is the smoothing coefficient, are the weights corresponding to different years, and ,in The attenuation factor takes a value between 0 and 1. is the current year, is the corresponding year, then predict the current year Hot selling time period The formula is ,in is the number of historical years; Compare the current time with the predicted hot-selling time interval of product A in the current year: If the current time is If the product A is in the hot-selling time range, If the current time is not range, then product A is in the cold-selling time period.

8. The cross-border e-commerce replenishment management system based on big data according to claim 1 is characterized by: The inventory monitoring and replenishment decision module (300) uses a line graph analysis method to analyze the daily sales quantity change trend predicted by the commodity sales volume prediction unit (220) within the current time interval; Draw a line graph with the date in the current time interval as the horizontal axis and the predicted daily sales quantity as the vertical axis. If the line graph is tilted upward, it is determined that the sales quantity is in an upward trend. If the line graph is tilted downward as a whole, it is determined that the sales quantity is in a downward trend. If the trend of change is in an upward trend, the date corresponding to the maximum sales volume is called out, the total sales volume of commodity A between the current date and the maximum sales volume date is calculated, it is set as the replenishment inventory quantity, and the replenishment inventory quantity signal is output at the same time.

9. The cross-border e-commerce replenishment management system based on big data according to claim 8 is characterized by: The sales forecast adjustment module (400) senses the current time node of commodity A, and adjusts the daily sales of commodity A in the corresponding previous time interval, and all daily sales in the previous time interval are sales set A, and divides the historical order data of commodity A into time intervals through the time interval division unit (210), and calls out the time interval corresponding to the previous time interval, which is defined as the matching time interval, and the daily sales of the commodity in each matching time interval is the sales set B; The similarity between the sales volume set A and the multiple sales volume sets B is matched and calculated, and the year corresponding to the sales volume set B with the maximum similarity is retrieved, and the year is defined as the first year. When the commodity sales volume prediction unit (220) uses the exponential smoothing method to predict the sales volume of commodity A every day in the current time interval, the weight of the previous year at the current time node is replaced with the weight of the first year, and the weight of the previous year is replaced with the weight of the year before that, until all years of commodity A are replaced.

Images