An automatic order pushing method and system based on fresh commodity sales prediction

Abstract

The application discloses an automatic order pushing method and system based on fresh commodity sales prediction, and relates to the field of supply chain management. The method is applied to a server, and the method comprises the following steps: obtaining sales data of multiple fresh stores in a target region; performing multi-time scale feature analysis on the sales data of the multiple fresh stores according to a preset integration dimension to obtain feature data sets of the multiple fresh stores at multiple time scales, wherein the preset integration dimension comprises a basic dimension and a derived dimension; inputting the feature data sets of the multiple fresh stores at the multiple time scales into a sales prediction model for cluster prediction to obtain sales prediction results of the multiple fresh stores; and generating comprehensive pushing orders according to the sales prediction results of the multiple fresh stores, and sending the comprehensive pushing orders to a supermarket order system. By implementing the technical scheme, the problems of low manual replenishment efficiency and low replenishment accuracy of fresh commodities in supermarkets are solved.

Description

Technical Field

[0001] This application relates to the technical field of supply chain management, specifically to an automated order push method and system based on sales forecasting of fresh produce. Background Technology

[0002] In the supermarket industry, fresh produce is difficult to replenish due to factors such as short shelf life, large fluctuations in demand, and significant seasonal influences, which greatly increases the operating costs of supermarkets.

[0003] Currently, to improve the accuracy of replenishment strategies, some automated order systems are used, which integrate sales forecasting algorithms for fresh produce to predict sales volume and provide a basis for replenishment strategy formulation. However, most of these sales forecasting algorithms use fixed rules or simple algorithms, treating market demand as having linear changes or stable periodicity. When the market environment undergoes irregular dynamic changes (promotional activities, shifts in consumer taste preferences), the accuracy of the sales forecast results will decrease. Summary of the Invention

[0004] In response to the problems of low efficiency and low accuracy of manual replenishment of fresh produce in supermarkets, this application provides an automated order push method and system based on sales forecasting of fresh produce.

[0005] Firstly, this application provides an automatic order push method based on fresh produce sales forecasting, applied to a server, the method comprising:

[0006] Obtain sales data from multiple fresh food stores in the target area;

[0007] Based on the preset integration dimensions, feature analysis is performed on the sales data of multiple fresh food stores at multiple time scales to obtain feature datasets of multiple fresh food stores at multiple time scales. The preset integration dimensions include basic dimensions and derived dimensions.

[0008] The feature datasets of multiple fresh food stores at multiple time scales are input into the sales prediction model for cluster prediction, and the sales prediction results of multiple fresh food stores are obtained.

[0009] Based on the sales forecasts of multiple fresh food stores, a comprehensive push order is generated and sent to the supermarket order system.

[0010] Optionally, obtaining sales data from multiple fresh food stores in the target area further includes:

[0011] The sales data of multiple fresh food stores are divided into real-time data, near real-time data, and static data, with the information update frequency decreasing in that order from real-time data to near real-time data to static data.

[0012] The unit access data is determined based on the ratio of the total amount of real-time data, near-real-time data, and static data. The unit access data consists of the real-time data, near-real-time data, and static data.

[0013] Based on the access data from the aforementioned units, sales data from multiple fresh food stores in the target region are obtained.

[0014] Optionally, determining the unit access data based on the ratio of the total data volume among the real-time data, near-real-time data, and static data further includes:

[0015] According to preset data association rules, the first data to be accessed is extracted from the real-time data, the second data to be accessed is extracted from the near-real-time data, and the third data to be accessed is extracted from the static data;

[0016] The first data to be accessed, the second data to be accessed, and the third data to be accessed are constructed into an associated data group;

[0017] The associated data group is identified as the unit access data.

[0018] Optionally, the step of performing multi-time-scale feature analysis on the sales data of multiple fresh food stores according to a preset integration dimension to obtain feature datasets of multiple fresh food stores at multiple time scales specifically includes:

[0019] Based on a preset minimum time scale selection range, multiple selectable minimum time scales are determined, wherein the multiple selectable minimum time scales are spaced at the same time interval.

[0020] Calculate the standard deviation of the sales data of the fresh food store to be analyzed at multiple selectable minimum time scales, wherein the fresh food store to be analyzed is any one of the multiple fresh food stores;

[0021] Based on the minimum time scale discriminant function, noise is judged on the standard deviation of the sales data of the fresh food store to be analyzed under multiple selectable minimum time scales, and the optimal selectable minimum time scale is determined.

[0022] Optionally, the step of determining multiple selectable minimum time scales based on the minimum time scale selection range further includes:

[0023] Based on the aforementioned basic dimensions, time intervals for multiple selectable minimum time scales are determined, including sales fluctuations, inventory turnover, and price fluctuations.

[0024] Based on the derived dimensions, the minimum time scale selection range is determined, and the derived dimensions include promotional conditions, weather conditions, and holidays.

[0025] Optionally, the minimum time scale discriminant function is specifically:

[0026]

[0027] in, The optimal choice is the smallest possible time scale. Let i be the i-th selectable minimum time scale. The threshold standard deviation, Let be the standard deviation of the sales data at the i-th selectable smallest time scale. The total number of time scales for sales data can be divided at the i-th smallest selectable time scale.

[0028] Optionally, the step of inputting feature datasets from multiple fresh food stores across multiple time scales into a sales prediction model for cluster prediction to obtain sales prediction results for the multiple fresh food stores specifically includes:

[0029] Clustering algorithms are used to divide the feature datasets of multiple fresh food stores across multiple time scales into clusters, resulting in multiple store clusters.

[0030] Machine learning models are trained on multiple store clusters to obtain cluster prediction models for multiple store clusters.

[0031] The sales forecast results for each of the aforementioned fresh food stores are obtained by inputting them into their respective cluster prediction models.

[0032] Secondly, this application provides an automatic order push system based on fresh produce sales forecasting. The system is a server, comprising a receiving module, a processing module, and a push module, wherein:

[0033] The receiving module is used to acquire sales data from multiple fresh food stores in the target area;

[0034] The processing module is used to perform multi-time-scale feature analysis on the sales data of multiple fresh food stores according to a preset integration dimension, to obtain feature datasets of multiple fresh food stores at multiple time scales, wherein the preset integration dimension includes a basic dimension and a derived dimension; and to input the feature datasets of multiple fresh food stores at multiple time scales into a sales prediction model for cluster prediction, to obtain the sales prediction results of multiple fresh food stores.

[0035] The push module is used to generate a comprehensive push order based on the sales forecast results of multiple fresh food stores, and send the comprehensive push order to the supermarket order system.

[0036] Thirdly, this application provides an electronic device including a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of the first aspects.

[0037] Fourthly, this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described in any one of the first aspects.

[0038] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0039] This application employs multi-timescale (minute / hour / day / week / month / quarter) feature analysis on real-time sales data from multiple fresh food stores, combining basic and derived dimensions to deeply uncover temporal patterns and potential influencing factors within the sales data. Then, through a cluster prediction mechanism, it accurately matches the sales models and market demand characteristics of different stores, achieving refined predictions of fresh food sales. Finally, based on the prediction results, it generates comprehensive push orders, ensuring a high degree of alignment between order replenishment and actual demand. When market demand fluctuates drastically, multi-timescale analysis can quickly capture short-term anomalies, allowing for timely adjustments to replenishment strategies and reducing the risk of stockouts. When demand stabilizes, long-term timescale feature analysis can accurately grasp sales trends, avoiding inventory buildup. This significantly improves the accuracy of fresh food sales forecasting, ensures the timeliness and stability of fresh food supply, and greatly enhances the operational efficiency and economic benefits of supermarket fresh food businesses. Attached Figure Description

[0040] Figure 1 This is a flowchart illustrating an automatic order push method based on sales forecasting of fresh produce, provided in an embodiment of this application.

[0041] Figure 2This is a schematic diagram of the structure of an automatic order push system based on sales forecasting of fresh produce, provided in an embodiment of this application.

[0042] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0043] Explanation of reference numerals in the attached drawings: 1. Receiving module; 2. Processing module; 3. Push module; 300. Electronic device; 301. Processor; 302. Communication bus; 303. User interface; 304. Network interface; 305. Memory. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0045] As the main supply node for many self-operated fresh food stores in the region, supermarkets' replenishment decisions directly determine the economic benefits of fresh food products in the region. However, in the supply chain management of fresh food products in the supermarket industry, replenishment decisions rely heavily on human experience, which presents many pain points.

[0046] Traditional methods, often based on simple historical sales averages or subjective judgments, struggle to accurately address the volatile demand, short shelf life, and significant seasonality of fresh produce. This leads to inventory buildup or stockouts, increasing losses and lost sales opportunities. Existing automated order systems often employ fixed rules or simple algorithms, treating market demand as linearly variable or exhibiting stable cyclicality. When the market environment undergoes irregular dynamic changes (promotional activities, shifts in consumer preferences), the accuracy of sales forecasts decreases, hindering the development of accurate replenishment decisions.

[0047] To address the aforementioned issues, this application provides an automatic order push method based on fresh produce sales forecasting. This method is applied to a server, such as... Figure 1 As shown, the method includes steps S101 to S104, which are as follows:

[0048] S101. Obtain sales data from multiple fresh food stores in the target area.

[0049] In the above steps, by connecting with the sales terminals of various fresh food stores, real-time sales records of each store are collected, including but not limited to sales quantity, sales time, product category, and price information. To ensure the comprehensiveness and accuracy of the data, inventory data and information related to promotional activities are also collected simultaneously. In addition, weather data and holiday information are accessed through external interfaces as an important supplement to subsequent multi-timescale feature analysis.

[0050] In one possible implementation, since the server accesses a wide variety of data and a large amount of data, but the server's information access frequency is limited, in order to improve the timeliness of the accessed information, this application divides the information to be accessed into real-time data, near-real-time data, and static data according to the information update frequency. The update frequency is from fast to slow as follows: real-time data, near-real-time data, and static data. Real-time data includes product sales records and inventory, near-real-time data includes information related to promotional activities and weather data, and static data includes store information and holiday information. Then, based on the ratio of the total amount of real-time data, near-real-time data, and static data, the unit access data at the current information access frequency is determined. For example, if the current information access frequency is 1M / s, and the ratio of the total amount of real-time data, near-real-time data, and static data is 7 / 2 / 1, then the unit access data consists of 0.7M real-time data, 0.2M near-real-time data, and 0.1M static data. Finally, based on the unit access data, the real-time data, near-real-time data, and static data are updated to the server's data pool. Compared to the polling access method, the above scheme avoids the situation where a large amount of certain data leads to a long access time, which in turn reduces the timeliness of subsequent access data.

[0051] Furthermore, during the information data access process, the composition of each accessed information is discrete. This leads to situations where the server cannot effectively utilize the accessed information as the access progresses, thus reducing the efficiency of subsequent information analysis. For example, when the information access progress reaches 50%, if the accessed information consists of "strawberry sales data records of store X, grape inventory data of store Y, and watermelon buy-two-get-one-free promotion information of store Z," for store X, although the complete strawberry sales data records have been accessed, the lack of strawberry inventory data and promotion information makes it difficult for the server to accurately predict its future sales. Effective prediction can only be made after the unaccessed data is fully accessed, thus reducing the efficiency of information analysis. Therefore, this application extracts the first data to be accessed from real-time data, the second data to be accessed from near-real-time data, and the third data to be accessed from static data according to preset data association rules. Then, the first, second, and third data to be accessed are constructed into an association data group. Finally, based on the association data group, each accessed information is associated and accessed, thereby achieving the effect of accessing and analyzing simultaneously, greatly improving the server's information analysis efficiency.

[0052] For missing or outlier values ​​appearing at the sales terminals of various fresh food stores, this application identifies whether the missing or outlier value is a key field. For key fields, the average value is taken and replaced according to the business logic. For example, if the sales volume of a certain fresh food product is missing on a certain day, the average value of the sales volume of the next day and the sales volume of the previous day is calculated and replaced. For non-key fields, this application directly sets the value to 0 or does not process them.

[0053] Finally, the accessed data is deduplicated and standardized in format to obtain the sales data of the fresh food stores.

[0054] S102. Based on the preset integration dimensions, perform feature analysis on the sales data of multiple fresh food stores at multiple time scales to obtain feature datasets of multiple fresh food stores at multiple time scales. The preset integration dimensions include basic dimensions and derived dimensions.

[0055] In the above steps, the basic dimensions can be understood as the explicit dimensions of sales data, which can be directly calculated from sales data. For example, the sales fluctuation index corresponding to sales fluctuation, the inventory turnover rate corresponding to inventory turnover, and the price fluctuation index corresponding to price fluctuation. The derived dimensions can be understood as the implicit dimensions of sales data, which require further data mining based on the basic dimensions. For example, the promotion sensitivity corresponding to promotion. The derived dimensions also include, but are not limited to, weather correlation and holiday effect coefficient.

[0056] Because the sales trend of fresh produce exhibits a complex and mixed characteristic, for example, it shows strong non-linear fluctuations in micro-trends (minutes / hours) and linear and periodic patterns in macro-trends (months / quarters), this application conducts feature analysis on multiple stores at multiple time scales to extract feature datasets for each store at multiple time scales. The feature datasets include multiple data corresponding to the basic dimensions and multiple data corresponding to the derived dimensions, thereby comprehensively mining the various characteristics of sales trend changes.

[0057] In one possible implementation, if the time scale is not chosen properly when mining various characteristics of sales trends, the mined characteristics may be greatly biased. For example, a chain supermarket records strawberry sales data at the minute level and finds that store A suddenly sold 50 boxes of strawberries in the minute between 10:15 and 10:16. If the time scale is chosen at the minute level, it may be concluded that "strawberry sales exploded at 10:15". However, the actual reason is that a corporate client made a bulk purchase for employee benefits in this minute, which is an isolated event. If a replenishment strategy is formulated based on the characteristics of this time scale, it may lead to overstocking, resulting in inventory backlog and losses. Therefore, to avoid this situation, this application first determines multiple selectable minimum time scales based on a preset minimum time scale selection range. These selectable minimum time scales are spaced at equal intervals; for example, if the minimum time scale selection range is 10 minutes to 4 hours, then the multiple selectable minimum time scales are 10 minutes, 20 minutes, 30 minutes...240 minutes, respectively. Then, the standard deviation of sales data under these multiple selectable minimum time scales is calculated. Finally, based on the minimum time scale discriminant function, the optimal selectable minimum time scale is determined. It should be explained that occasional events can be considered as noise; therefore, the optimal selectable minimum time scale is the time scale where the sales trend is least affected by noise. The minimum time scale discriminant function is as follows:

[0058]

[0059] in, The optimal choice is the smallest possible time scale. Let i be the i-th selectable minimum time scale. The threshold standard deviation, Let be the standard deviation of the sales data at the i-th selectable smallest time scale. The total number of time scales for sales data can be divided at the i-th smallest selectable time scale.

[0060] In the above formula, the threshold standard deviation This can be understood as the critical value between the effective trend of sales data and random noise. When the threshold standard deviation is exceeded, the sales trend is dominated by noise. In order to reasonably set the threshold standard deviation, this application, based on the statistical principle of normal distribution, selects twice the standard deviation calculated from the lower limit of the minimum time scale selection range (2). Using the high confidence interval as the threshold standard deviation, noise is filtered out while retaining valid data; when Much larger This indicates that the data at this time scale is noisy and fluctuates wildly, failing to effectively reflect the trend. much smaller This indicates excessive smoothing at that time scale, which may cause the loss of true business fluctuations; therefore, when and The closer the features are, the better they can filter out noise while preserving effective characteristics; finally, the objective function... Calculate The minimum selectable time scale corresponding to the minimum value is taken as the optimal selectable minimum time scale.

[0061] In one possible implementation, to make the calculation results of the preset minimum time scale discriminant function more closely match the sales scenario and sales situation of fresh food stores, this application determines multiple selectable minimum time scale time intervals based on the basic dimensions. Specifically, it identifies the change cycles of multiple data in the basic dimensions, and then uses the time interval corresponding to the minimum change cycle as the time interval of multiple selectable minimum time scales. For example, if sales fluctuate at least once every 5 minutes, inventory turns over at least once every half hour, and prices change at least once every 2 hours, then the minimum change cycle is once every 5 minutes. In this case, the time interval of multiple selectable minimum time scales is 5 minutes, thus aligning with the minimum sales situation. Effective granularity is determined first; then, based on the derived dimensions, the minimum time scale selection range is determined. Specifically, the effective impact periods of multiple data points within the derived dimensions are identified. The time interval corresponding to the smallest effective impact period is used as the lower limit of the minimum time scale selection range, and the time interval corresponding to the largest effective impact period is used as the upper limit. For example, if the effective time for a promotional activity is 2 hours, the effective impact time for weather is 12 hours, and the effective impact time for holidays is 72 hours, then the smallest effective impact period is 2 hours, and the largest effective impact period is 72 hours. Therefore, the minimum time scale selection range is from 2 hours to 72 hours, thus ensuring full coverage of the sales scenario. In the above method, the basic and derived dimensions are deeply bound to the time scale selection, ensuring that the optimal selectable minimum time scale both fits the data and adapts to the sales business.

[0062] S103. Input the feature datasets of multiple fresh food stores at multiple time scales into the sales prediction model for cluster prediction, and obtain the sales prediction results of multiple fresh food stores.

[0063] In the aforementioned steps, before predicting sales for multiple fresh food stores, the accuracy of the constructed sales prediction model is low due to the diverse product categories, frequent updates to sales data, and significant differences in sales models among each store. This application employs a clustering prediction strategy. First, a clustering algorithm is used to divide the feature datasets of multiple fresh food stores across multiple time scales, resulting in multiple store clusters. Specifically, this is achieved by quantifying the feature similarity between different stores. For example, if two stores have similar feature values ​​across multiple dimensions such as sales trend, inventory turnover rate, and promotional sensitivity, they will be grouped into the same group. After clustering, machine learning models are trained on each of the multiple store clusters to obtain cluster prediction models. Finally, each fresh food store is input into its corresponding cluster prediction model to obtain the sales prediction results for multiple fresh food stores.

[0064] S104. Based on the sales forecast results of multiple fresh food stores, generate a comprehensive push order and send the comprehensive push order to the supermarket order system.

[0065] In the above steps, the sales forecast results include the predicted sales volume of fresh food stores for the next few days. Therefore, the replenishment quantity for stores with next-day delivery = (today's predicted sales volume + tomorrow's predicted sales volume) * (1 + loss rate) + minimum shelf space - today's initial inventory. The replenishment quantity for stores with two-day delivery = (today's predicted sales volume + tomorrow's predicted sales volume + the day after tomorrow's predicted sales volume) * (1 + loss rate) + minimum shelf space - today's initial inventory. The minimum shelf space is the minimum quantity that needs to be displayed for sale on the shelves of the fresh food store. The above replenishment quantities are all replenishment data for a single category of fresh food, and the loss rate is the loss rate of that category of fresh food after the end of the day's sales. Finally, the replenishment quantities of each fresh food store are generated into a comprehensive push order, which is then sent to the supermarket order system. Supermarket staff purchase goods in advance based on the comprehensive push order, preparing the goods to be supplied before each store starts submitting orders each day, thereby meeting the sales needs of each store.

[0066] In one possible implementation, after the supermarket order system receives the daily orders from each store, it feeds the daily orders from each store back to the server. The server calculates the prediction deviation for each store based on the sales forecast results of multiple fresh food stores and the daily orders from each store. Then, based on the prediction error of each store, the model parameters of the sales forecast model are corrected, which serves as a reference for subsequent sales forecast models, thereby improving the accuracy of the sales forecast results.

[0067] Reference Figure 2 This application also provides an automatic order push system based on fresh produce sales forecasting. The system is a server, including a receiving module 1, a processing module 2, and a push module 3, wherein:

[0068] Module 1 is used to acquire sales data from multiple fresh food stores in the target area.

[0069] Processing module 2 is used to perform multi-time-scale feature analysis on the sales data of multiple fresh food stores according to the preset integration dimensions, to obtain feature datasets of multiple fresh food stores at multiple time scales. The preset integration dimensions include basic dimensions and derived dimensions. The feature datasets of multiple fresh food stores at multiple time scales are input into the sales prediction model for cluster prediction to obtain the sales prediction results of multiple fresh food stores.

[0070] Push module 3 is used to generate a comprehensive push order based on the sales forecast results of multiple fresh food stores, and send the comprehensive push order to the supermarket order system.

[0071] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0072] This application also discloses an electronic device. (See reference...) Figure 3 , Figure 3 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. The electronic device 300 may include: at least one processor 301, at least one network interface 304, a user interface 303, a memory 305, and at least one communication bus 302.

[0073] The communication bus 302 is used to enable communication between these components.

[0074] The user interface 303 may include a display screen and a camera. Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0075] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0076] The processor 301 may include one or more processing cores. The processor 301 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 305, and by calling data stored in memory 305. Optionally, the processor 301 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 301 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 301 and may be implemented as a separate chip.

[0077] The memory 305 may include random access memory (RAM) or read-only memory. Optionally, the memory 305 may include a non-transitory computer-readable storage medium. The memory 305 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 305 may also be at least one storage device located remotely from the aforementioned processor 301. (Refer to...) Figure 3 The memory 305, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for an automatic order push method based on sales forecasting of fresh produce.

[0078] exist Figure 3In the illustrated electronic device 300, the user interface 303 is mainly used to provide an input interface for the user and to acquire user input data; while the processor 301 can be used to call an application stored in the memory 305 for an automatic order push method based on fresh produce sales forecasting. When executed by one or more processors 301, the electronic device 300 performs one or more of the methods described in the above embodiments. It should be noted that, for the foregoing method embodiments, for the sake of simplicity, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0079] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0080] In the various embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.

[0081] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0082] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0083] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0084] The above description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will be readily apparent to those skilled in the art upon consideration of the specification and the disclosure of practical truths.

[0085] This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. An automatic order push method based on fresh produce sales forecasting, characterized in that, Applied to a server, the method includes: Obtain sales data from multiple fresh food stores in the target area; Based on preset integration dimensions, multi-time-scale feature analysis is performed on the sales data of multiple fresh food stores to obtain feature datasets of the multiple fresh food stores at multiple time scales. The preset integration dimensions include basic dimensions and derived dimensions, specifically including: Based on the minimum time scale selection range, multiple selectable minimum time scales are determined, wherein the multiple selectable minimum time scales are spaced at the same time interval, which specifically includes: Based on the aforementioned basic dimensions, a number of selectable minimum time scales are determined. The basic dimensions include sales fluctuations, inventory turnover, and price fluctuations. Specifically, the change cycles of multiple data in the basic dimensions are identified, and the time interval corresponding to the smallest change cycle is taken as the time interval of the multiple selectable minimum time scales. Based on the derived dimensions, the minimum time scale selection range is determined. The derived dimensions include promotional conditions, weather conditions, and holidays. Specifically, the effective influence periods of multiple data in the derived dimensions are identified. The time interval corresponding to the smallest effective influence period is taken as the lower limit of the minimum time scale selection range, and the time interval corresponding to the largest effective influence period is taken as the upper limit of the minimum time scale selection range. Calculate the standard deviation of the sales data of the fresh food store to be analyzed at multiple selectable minimum time scales, wherein the fresh food store to be analyzed is any one of the multiple fresh food stores; Based on the minimum time scale discriminant function, noise is assessed on the standard deviation of the sales data of the fresh food store to be analyzed under multiple selectable minimum time scales to determine the optimal selectable minimum time scale. The minimum time scale discriminant function is specifically: in, The optimal choice is the smallest possible time scale. Let i be the i-th selectable minimum time scale. The threshold standard deviation, Let be the standard deviation of the sales data at the i-th selectable smallest time scale. The total number of time scales that can be divided at the i-th smallest selectable time scale for sales data; The feature datasets of multiple fresh food stores at multiple time scales are input into the sales prediction model for cluster prediction, and the sales prediction results of multiple fresh food stores are obtained. Based on the sales forecasts of multiple fresh food stores, a comprehensive push order is generated and sent to the supermarket order system.

2. The method according to claim 1, characterized in that, The acquisition of sales data from multiple fresh food stores in the target area specifically includes: The sales data of multiple fresh food stores are divided into real-time data, near real-time data, and static data, with the information update frequency decreasing in that order from real-time data to near real-time data to static data. The unit access data is determined based on the ratio of the total amount of real-time data, near-real-time data, and static data. The unit access data consists of the real-time data, near-real-time data, and static data. Based on the access data from the aforementioned units, sales data from multiple fresh food stores in the target region are obtained.

3. The method according to claim 2, characterized in that, The step of determining the unit access data based on the ratio of the total data volume among the real-time data, near-real-time data, and static data further includes: According to preset data association rules, the first data to be accessed is extracted from the real-time data, the second data to be accessed is extracted from the near-real-time data, and the third data to be accessed is extracted from the static data; The first data to be accessed, the second data to be accessed, and the third data to be accessed are constructed into an associated data group; The associated data group is identified as the unit access data.

4. The method according to claim 1, characterized in that, The step of inputting feature datasets from multiple fresh food stores across multiple time scales into a sales prediction model for cluster prediction, and obtaining sales prediction results for the multiple fresh food stores, specifically includes: Clustering algorithms are used to divide the feature datasets of multiple fresh food stores across multiple time scales into clusters, resulting in multiple store clusters. Machine learning models are trained on multiple store clusters to obtain cluster prediction models for multiple store clusters. The sales forecast results for each of the aforementioned fresh food stores are obtained by inputting them into their respective cluster prediction models.

5. An automatic order push system based on fresh produce sales forecasting, the system being used to execute an automatic order push method based on fresh produce sales forecasting as described in any one of claims 1-4, characterized in that, The system is a server, including a receiving module (1), a processing module (2), and a pushing module (3), wherein: The receiving module (1) is used to obtain sales data from multiple fresh food stores in the target area; The processing module (2) is used to perform multi-time-scale feature analysis on the sales data of multiple fresh food stores according to a preset integration dimension, to obtain feature datasets of multiple fresh food stores at multiple time scales, wherein the preset integration dimension includes a basic dimension and a derived dimension; and input the feature datasets of multiple fresh food stores at multiple time scales into the sales prediction model for cluster prediction, to obtain the sales prediction results of multiple fresh food stores. The push module (3) is used to generate a comprehensive push order based on the sales forecast results of multiple fresh food stores, and send the comprehensive push order to the supermarket order system.

6. An electronic device, characterized in that, The device includes a processor (301), a memory (305), a user interface (303), and a network interface (304). The memory (305) is used to store instructions. The user interface (303) and the network interface (304) are used to communicate with other devices. The processor (301) is used to execute the instructions stored in the memory (305) to cause the electronic device (300) to perform the method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1 to 4.

Images