CN114881694A

CN114881694A - Automatic replenishment method, system, electronic device, storage medium, and program product

Info

Publication number: CN114881694A
Application number: CN202210507362.4A
Authority: CN
Inventors: 邱克; 吴清灏; 高雄; 覃志强; 姚仲南; 孙博洋
Original assignee: Beijing Shiji Dashang Information Technology Co ltd
Current assignee: Beijing Shiji Dashang Information Technology Co ltd
Priority date: 2022-05-10
Filing date: 2022-05-10
Publication date: 2022-08-09

Abstract

The invention provides an automatic replenishment method, an automatic replenishment system, electronic equipment, a storage medium and a program product, wherein the method comprises the following steps: obtaining a sales prediction result based on historical sales data through a sales prediction model; determining replenishment information based on the sales forecast result; wherein the obtaining of the sales prediction model comprises: extracting data characteristics of commodities to be restocked based on the historical sales data of each commodity, classifying all the commodities based on all the data characteristics, and matching preset sales prediction models for different classifications. By automatically extracting the data characteristics of the commodities, the optimal sales prediction model suitable for the prediction object is matched, manual experience is not relied on, and the accuracy and the automation degree of the prediction model are improved.

Description

Automatic replenishment method, system, electronic device, storage medium, and program product

Technical Field

The present invention relates to the field of business forecasting technologies, and in particular, to an automatic replenishment method, system, electronic device, storage medium, and program product.

Background

Replenishment is one of the most core businesses of retailers, and various replenishment methods are continuously explored and practiced in the development process of the retail industry so as to obtain the best benefit between the guarantee of sales and the effective reduction of purchasing cost. The accurate prediction of the commodity sales number is a key factor in the replenishment process, and the replenishment number also needs to comprehensively consider the factors of logistics, storage, supplier cooperation and the like.

Currently, retailers use several methods for restocking:

1) the expert method comprises the following steps: the business personnel can determine what to supplement and how much to supplement according to personal experience and simple sales data. The method effectively makes a decision by means of abundant experience of business personnel and comprehensively considering the change of market factors, but is difficult to make correct judgment under the background that the current commodity variety is more and the updating speed is accelerated because of the inevitable rigor of personal experience, and the method cannot support automatic replenishment of the system.

2) Moving average method: and taking the average sales volume of a plurality of past periods as the future demand volume, and performing weight adjustment to perform replenishment. The method is simple and efficient in calculation, and has good applicability to commodities with stable distribution and sale conditions. But the weight adjustment is heavily dependent on personal experience and the high frequency of weight adjustment is difficult to implement in business practice.

3) Stock water level method: and setting a maximum inventory and a minimum inventory for the goods, and supplementing the maximum inventory when the actual inventory is lower than the minimum inventory. The system is simple and efficient, is mostly applied to an automatic replenishment system, but the setting and adjustment of the minimum and maximum inventory are very dependent on manual experience, and are difficult to adjust in time.

4) An artificial intelligence replenishment method: the sales quantity is predicted mainly in a regression or time series mode, and automatic replenishment calculation is carried out. Because retail industry data is huge and AI calculation is complicated, a single prediction model is adopted for calculation at present.

In the retail industry, replenishment objects are thousands of commodities, the retail transaction randomness is high, the season difference is obvious, the data volume is large, and the timeliness requirement is high. In the prior art, predictive modeling has low automation degree, more links depending on manual experience processing, large prediction workload, low efficiency and poor precision.

Disclosure of Invention

The invention provides an automatic replenishment method, an automatic replenishment system, electronic equipment, a storage medium and a program product, aiming at the problems in the prior art.

The invention provides an automatic replenishment method, which comprises the following steps:

obtaining a sales prediction result based on historical sales data through a sales prediction model;

determining replenishment information based on the sales forecast result;

wherein the obtaining of the sales prediction model comprises: extracting data characteristics of commodities to be restocked based on the historical sales data of each commodity, classifying all the commodities based on all the data characteristics, and matching preset sales prediction models for different classifications.

According to the automatic replenishment method provided by the invention, the step of classifying all the commodities based on all the data characteristics comprises the following steps:

and classifying all the commodities by a clustering method based on all the data characteristics.

According to the automatic replenishment method provided by the invention, the data characteristics comprise the following two items:

discrete coefficients, periodic coefficients.

According to the automatic replenishment method provided by the invention, the clustering method adopts a K-means clustering algorithm, and the commodities are divided into four categories;

matching competition models for the two types of commodities with smaller discrete coefficients; matching a regression prediction model for the two types of commodities with the larger discrete coefficients and the periodic coefficients larger than zero;

and matching a moving average model for the two types of commodities with the larger discrete coefficients and the periodic coefficient less than or equal to zero.

According to the automatic replenishment method provided by the invention, for the two types of commodities with smaller discrete coefficients, a competition model is matched, and the method comprises the following steps:

adopting a plurality of sub-prediction models preset in the competition model to respectively obtain a plurality of corresponding prediction results;

and selecting the sub-prediction model which is most matched with the commodity as the competition model based on the weighted average absolute percentage error of each of the plurality of prediction results.

According to the automatic replenishment method provided by the invention, the method further comprises the following steps:

and if the sales prediction model matched with the commodity is the competition model or the regression prediction model, predicting by adopting a preset parameter combination on the basis of the competition model or the regression prediction model, and selecting an optimal parameter combination of each commodity by using K-fold cross validation.

periodically calculating the weighted average absolute percentage error of the prediction result of each commodity in the period;

and for the commodities with the weighted average absolute percentage error which is reduced to exceed the set threshold value, acquiring the sales prediction model again.

According to the automatic replenishment method provided by the invention, the determination of the replenishment information based on the sales prediction result comprises the following steps:

forming a quartile frame body by counting the quartile of the sales prediction result;

and replacing the prediction result outside the quartet frame by using a frame upper limit value and a frame lower limit value.

According to the automatic replenishment method provided by the invention, the data characteristics of the commodities to be replenished are extracted based on the historical sales data of each commodity, and the method comprises the following steps:

for the same kind of the commodity with a plurality of sales subjects, a plurality of data features are extracted based on the historical sales data of different sales subjects.

The invention also provides an automatic replenishment system, which comprises:

the prediction module acquires a sales prediction result based on historical sales data through a sales prediction model;

a replenishment module that determines replenishment information based on the sales prediction result;

The invention also provides an electronic device, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the steps of the automatic replenishment method.

The invention also provides a non-transitory computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the method of automatic restocking as described in any of the above.

The invention also provides a computer program product comprising a computer program which, when executed by a processor, carries out the steps of the method of automatic restocking as described in any of the above.

According to the automatic replenishment method, the automatic replenishment system, the electronic equipment, the storage medium and the program product, the data characteristics of the commodities are automatically extracted, the optimal sales prediction model suitable for the prediction object is matched, manual experience is not relied on, and the accuracy and the automation degree of the prediction model are improved.

Drawings

In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

FIG. 1 is a schematic flow chart of an automatic replenishment method according to the present invention;

FIG. 2 is a schematic structural diagram of an automatic replenishment system according to the present invention;

fig. 3 is a schematic physical structure diagram of an electronic device according to the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The following describes in detail the automatic replenishment method provided in the embodiment of the present application with reference to the drawings and a specific embodiment and an application scenario thereof.

Fig. 1 is a schematic flow chart of an automatic replenishment method provided by the present invention, and as shown in fig. 1, the automatic replenishment method provided by the present invention includes:

s100, obtaining a sales prediction result based on historical sales data through a sales prediction model;

s200, determining replenishment information based on a sales prediction result; wherein the obtaining of the sales prediction model comprises: and extracting data characteristics of commodities to be restocked based on historical sales data of each commodity, classifying all commodities based on all the data characteristics, and matching a preset sales prediction model according to different classifications.

It should be noted that, with the iterative update of the historical sales data, if the data characteristics corresponding to each commodity change, the classification situation may change accordingly, and the method of this embodiment enables the sales prediction model matched therewith to automatically change accordingly, thereby implementing highly automated model matching.

Optionally, the historical sales data is historical sales data over two years.

Optionally, the sales prediction result comprises a prediction of sales volume 7 days after the prediction time.

Optionally, preset sales prediction models are matched according to different categories, for each commodity, the parameter most matched with the corresponding sales prediction model is selected, and the commodity-sales prediction model-parameter association is stored in the commodity mode list.

Optionally, the sales data of the previous day is imported into the sales prediction model every day, and the corresponding sales prediction model and parameters are obtained from the commodity pattern list according to the commodity type, so as to predict the sales quantity of the next 7 days.

Alternatively, the replenishment quantity of each commodity is calculated from the sales prediction result, the stock quantity, the order quantity, and the replenishment parameter (the number of days to arrive, the minimum order quantity, etc.).

Further, a replenishment triggering time point algorithm is provided: (quantity in stock + quantity ordered in transit)/forecast daily average sales < (target days in stock + days to delivery);

further, a restocking quantity algorithm is provided: (target days of inventory + days to goods one ((quantity of inventory-on order quantity)/forecast average daily sales)). forecast average daily sales;

further, an actual replenishment quantity algorithm is provided: and (4) calculating by integrating related parameters of different suppliers of each commodity, including the minimum shipment volume, the piece number and the delivery calendar.

Optionally, indexes such as the number of days of stock turnover, a sold-out ratio, a stock ratio and a sold ratio, a moving sales ratio, a goods outage rate, a stock sales ratio and the like are adopted to help customers to track the goods replenishment effect and improve the shortage of the service management process.

Further, after determining the specific replenishment quantity, the replenishment quantity is input into an ERP (Enterprise Resource Planning) order system.

According to the method and the device, the data characteristics of the commodities are automatically extracted, the optimal sales prediction model suitable for the prediction object is matched, manual experience is not relied on, and the accuracy and the automation degree of the sales prediction model are improved.

Further, on the basis of the foregoing embodiment, in another embodiment, the present embodiment provides an automatic replenishment method for classifying all commodities based on all data features, including:

and classifying all commodities by a clustering method based on all data characteristics.

It should be noted that the clustering method is an unsupervised learning method, and the automatic classification of all the commodities is realized by the clustering method, and the method brings beneficial effects, including: with the change of the historical sales data of the commodities and the increase or decrease of the commodities, all the commodities can be automatically reclassified, the automation degree is high, manual classification is not relied on, and a proper sales prediction model can be better matched.

According to the method, all commodities are classified through a clustering method based on all data characteristics, the accuracy of matching the sales prediction model is improved, and the automation degree of the whole method is also improved.

Further, on the basis of the foregoing embodiment, in another embodiment, the present embodiment provides an automatic replenishment method, and the data features include the following two items:

discrete coefficients, periodic coefficients.

Alternatively, the periodic coefficient is a correlation coefficient value calculated using a ternary linearity (a periodic coefficient (1-7), a daily average sales amount of the last 2 days, a daily average sales amount of the last 5 days).

The embodiment discloses a specific data feature acquisition method, which completes index construction for subsequent classification based on the data feature.

Further, on the basis of the foregoing embodiment, in another embodiment, the present embodiment provides an automatic replenishment method, in which a K-means clustering algorithm is adopted in the clustering method, and the commodities are classified into four categories;

for two types of commodities with smaller discrete coefficients, matching competition models; matching a regression prediction model for two types of commodities with larger discrete coefficients and cycle coefficients larger than zero;

and matching the moving average model for the two types of commodities with larger discrete coefficients and the periodic coefficient less than or equal to zero.

It should be noted that, in the present embodiment, K is 4, which is a K-means clustering algorithm.

Optionally, the establishing of the regression prediction model comprises: and calculating the correlation coefficient of the data characteristics through linear regression analysis according to the historical sales data, so as to obtain a multiple regression prediction model as a baseline model.

Optionally, the moving average model specifically adopts a 21-day moving average prediction algorithm.

The embodiment discloses a specific classification method, and through the classification scheme, the sales prediction model matched with the commodities has better prediction accuracy.

Further, on the basis of the foregoing embodiment, in another embodiment, the present embodiment provides an automatic replenishment method for matching a competitive model for two types of commodities with smaller discrete coefficients, including:

adopting a plurality of sub-prediction models preset in a competition model to respectively obtain a plurality of corresponding prediction results;

and selecting the sub-prediction model with the best matching commodity as a competition model based on the weighted average absolute percentage error of each of the plurality of prediction results.

It should be noted that the weighted Mean Absolute Percentage error described in the present invention, namely wmape (weighted Mean Absolute Percentage error).

Optionally, the plurality of sub-prediction models specifically include: ARIMA, Prophet. Furthermore, ARIMA and Prophet are used for respectively carrying out prediction operation on 9-24 months of historical sales data, and then a WMAPE index is used for judging which sub-prediction model is more excellent.

The embodiment enables better prediction accuracy on the commodities with smaller discrete coefficients through the competition model.

Further, on the basis of the foregoing embodiment, in another embodiment, the present embodiment provides an automatic replenishment method, and the method further includes:

if the sales prediction model matched with the commodities is a competition model or a regression prediction model, adopting a preset parameter combination for prediction on the basis of the competition model or the regression prediction model, and selecting an optimal parameter combination of each commodity by using K-fold cross validation.

It should be noted that the preset parameter combinations refer to that some parameter combinations are set for training the model according to the results of the earlier research under the limitation of computing power and timeliness, so as to reduce the computing power requirement and ensure the timeliness of the training.

Optionally, after determining the optimal parameter combination, storing the commodity-sales prediction model-optimal parameter combination association in a commodity pattern list.

According to the embodiment, the calculation force requirement is reduced and the training timeliness is guaranteed through the preset parameter combination training model, meanwhile, the optimal parameter combination of each commodity is selected through K-fold cross validation, and the prediction accuracy is improved.

and for the commodities with the weighted average absolute percentage error which is reduced to exceed the set threshold, acquiring the sales prediction model again.

Optionally, the period is one month.

Further, the method of the embodiment is executed at the beginning of each month, and the acquisition of the sales prediction model is performed again for the commodities with the WMAPE ring ratio decreasing and exceeding the set threshold.

According to the embodiment, the method or the system applying the method can automatically and periodically iterate and optimize parameters, and the prediction accuracy is kept.

Further, on the basis of the foregoing embodiment, in another embodiment, the present embodiment provides an automatic replenishment method for determining replenishment information based on a sales prediction result, where the method includes:

forming a four-quadrant frame body by counting quartiles of a sales prediction result;

and replacing the prediction result outside the quartet frame by using the upper and lower frame limit values.

It should be noted that accidental data problems in engineering practice may lead to significant and unreasonable sales prediction results, and the embodiment uses a quarter frame to replace the prediction results outside the frame with frame upper and lower limit values (i.e., Whisker upper limit and Whisker lower limit) to correct these extreme values.

According to the method and the device, unreasonable predicted values caused by data abnormity are avoided through an outlier frame limiting technology, so that abnormal replenishment suggestions are generated, the prediction precision is improved, and more reasonable replenishment information is provided.

Further, on the basis of the foregoing embodiment, in another embodiment, the present embodiment provides an automatic replenishment method, for a commodity to be replenished, extracting data features based on historical sales data of each commodity, including:

for the same kind of merchandise with a plurality of sales subjects, a plurality of data features are extracted based on historical sales data of different sales subjects.

Alternatively, the selling body is a supermarket, a shop, an automatic vending machine, or the like.

It should be noted that, different sales subjects cause different sales geographical locations and sales scenes, and the sales prediction models matched with the same kind of goods may be greatly different, so for the same kind of goods but different sales subjects, the sales prediction models are regarded as different kinds of goods, and data features, clusters, and matching are respectively extracted.

Optionally, the sales data of the previous day is imported into the sales prediction model every day, and according to the commodity ID and the store ID, the corresponding sales prediction model and parameters are obtained from the commodity pattern list, and the sales volume prediction of the next 7 days is performed.

According to the method and the device, the commodity types participating in prediction are divided based on the sales subject, so that the same commodity of different sales subjects is matched with different sales prediction models, and prediction is more accurate.

The following describes the automatic replenishment system provided by the present invention, and the automatic replenishment system described below and the automatic replenishment method described above may be referred to in correspondence with each other.

Fig. 2 is a schematic structural diagram of an automatic replenishment system provided by the present invention, and as shown in fig. 2, the present invention further provides an automatic replenishment system, which includes:

the replenishment module determines replenishment information based on the sales prediction result;

wherein the obtaining of the sales prediction model comprises: and extracting data characteristics of commodities to be restocked based on historical sales data of each commodity, classifying all commodities based on all the data characteristics, and matching a preset sales prediction model according to different classifications.

According to the method and the device, the data characteristics of the commodities are automatically extracted, the optimal sales prediction model suitable for the prediction object is matched, and manual experience is not relied on.

Furthermore, the automatic replenishment system introduces the automatic replenishment method, improves the accuracy and the degree of automation of the prediction model, provides a set of efficient and complete automatic replenishment calculation engine to solve the problem that the accuracy of the replenishment quantity is influenced due to poor sales prediction precision in the replenishment quantity calculation, and also provides a report tool to track the replenishment effect and help business departments to improve the replenishment management efficiency.

Fig. 3 is a schematic physical structure diagram of an electronic device provided in the present invention, and as shown in fig. 3, the electronic device may include: a processor (processor)810, a communication Interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication Interface 820 and the memory 830 communicate with each other via the communication bus 840. The processor 810 may call logic instructions in the memory 830 to perform an auto-replenishment method, the method comprising:

determining replenishment information based on the sales forecast result;

In addition, the logic instructions in the memory 830 can be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

In another aspect, the present invention also provides a computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform the automatic replenishment method provided by the above methods, the method comprising:

determining replenishment information based on the sales forecast result;

In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor is implemented to perform the automatic replenishment methods provided above, the method comprising:

determining replenishment information based on the sales forecast result;

The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.

Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.

Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims

1. An automatic restocking method, characterized in that the method comprises:

determining replenishment information based on the sales forecast result;

2. The method of claim 1, wherein said sorting all of said items based on all of said data characteristics comprises:

3. The method of claim 2, wherein the data characteristics include two of:

discrete coefficients, periodic coefficients.

4. The automatic restocking method according to claim 3, wherein the clustering method employs a K-means clustering algorithm and classifies the commodities into four categories;

5. The automatic restocking method according to claim 4, wherein said matching a competition model for the two types of said commodities whose dispersion coefficients are small comprises:

6. The method of automatic restocking according to claim 4, wherein said method further comprises:

7. The automatic restocking method according to any one of claims 1-6, wherein the method further comprises:

8. The automatic restocking method according to claim 1, wherein said determining restocking information based on said sales prediction result previously comprises:

9. The automatic restocking method according to any one of claims 1, wherein said extracting data features based on the historical sales data of each of the commodities for the commodity to be restocked, comprises:

10. An automatic restocking system, the system comprising:

11. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program implements the steps of the method of automatic replenishment as claimed in any one of claims 1-9.

12. A non-transitory computer readable storage medium, having stored thereon a computer program, wherein the computer program, when executed by a processor, performs the steps of the method for automatic restocking according to any one of claims 1-9.

13. A computer program product comprising a computer program, wherein the computer program when executed by a processor performs the steps of the method for automatic restocking as claimed in any one of claims 1-9.