CN114626896A

CN114626896A - Method and device for predicting quantity of articles, electronic equipment and storage medium

Info

Publication number: CN114626896A
Application number: CN202210351747.6A
Authority: CN
Inventors: 张轩琪
Original assignee: Beijing Jingdong Zhenshi Information Technology Co Ltd
Current assignee: Beijing Jingdong Zhenshi Information Technology Co Ltd
Priority date: 2022-04-02
Filing date: 2022-04-02
Publication date: 2022-06-14

Abstract

The disclosure relates to an article quantity prediction method and device, electronic equipment and a computer readable storage medium, relates to the technical field of computers, and can be applied to a scene of sales prediction of a certain article. The method comprises the following steps: acquiring historical service time sequence data of a target article, and performing time sequence feature extraction processing on the historical service time sequence data to determine time prediction granularity corresponding to the target article; respectively carrying out regional aggregation prediction processing and regional disassembly prediction processing on the historical service time sequence data according to the time prediction granularity so as to determine the spatial prediction granularity corresponding to the target object; and determining the quantity of the target object in the specific business according to the time prediction granularity and the space prediction granularity. The present disclosure may provide a scheme that combines temporal and spatial dimensions to derive a prediction dimension for the item, by providing a configuration that is accurate to the item level, to derive an item quantity prediction result that integrates temporal and spatial dimensions.

Description

Method and device for predicting quantity of articles, electronic equipment and storage medium

Technical Field

The present disclosure relates to the field of computer technologies, and in particular, to an item quantity prediction method, an item quantity prediction apparatus, an electronic device, and a computer-readable storage medium.

Background

Sales forecast refers to forecast of future sales based on past sales and using sales forecasting models built into the system or customized by the user. The sales forecast comprises a forecast of sales of different items. In the process of selling goods, the rapid increase of sales volume caused by shopping festival events or other promotion events poses great challenges to retailers and can cause the problems of out-of-stock or system breakdown and the like, so that the prediction of the sales volume is crucial to the optimization of supply chains, the reduction of operation cost and the improvement of income of online and online retail industries.

It is noted that the information disclosed in the above background section is only for enhancement of understanding of the background of the present disclosure and therefore may include information that does not constitute prior art that is already known to a person of ordinary skill in the art.

Disclosure of Invention

The present disclosure is directed to an item quantity prediction method, an item quantity prediction apparatus, an electronic device, and a computer-readable storage medium, so as to overcome, at least to some extent, a problem that in the prior art, sales of an item is usually predicted only for a time dimension or a space dimension, and a prediction result cannot be determined by integrating the time dimension and the space dimension.

Additional features and advantages of the disclosure will be set forth in the detailed description which follows, or in part will be obvious from the description, or may be learned by practice of the invention.

According to a first aspect of the present disclosure, there is provided an item quantity prediction method, including: acquiring historical service time sequence data of a target article, and performing time sequence feature extraction processing on the historical service time sequence data to determine a time prediction granularity corresponding to the target article; respectively carrying out regional aggregation prediction processing and regional disassembly prediction processing on the historical service time sequence data according to the time prediction granularity so as to determine the spatial prediction granularity corresponding to the target article; and determining the item quantity of the target item in a specific service according to the temporal prediction granularity and the spatial prediction granularity.

In an exemplary embodiment of the present disclosure, performing a time series feature extraction process on the historical traffic time series data to determine a time prediction granularity corresponding to the target item includes: acquiring a pre-constructed time sequence analysis model; the time sequence analysis model is obtained based on a plurality of time sequence division data training; inputting the historical traffic timing data to the timing analysis model to determine the temporal prediction granularity from the timing analysis model; the time sequence analysis model is obtained by training the following steps: acquiring an initial time sequence model, a service time sequence training data set and a plurality of pre-configured candidate time granularities; dividing the service time sequence training data set according to the candidate time granularities to obtain a plurality of time sequence division data; and carrying out model training processing on the initial time sequence model according to the plurality of time sequence division data to obtain the time sequence analysis model.

In an exemplary embodiment of the present disclosure, performing a region aggregation prediction process and a region disassembly prediction process on the historical service time series data according to the time prediction granularity, respectively, to determine a spatial prediction granularity corresponding to the target item, includes: determining area level information corresponding to the target object; performing region aggregation prediction processing on the historical service time sequence data according to the region level information to obtain a first accuracy result corresponding to a first space division granularity; performing region disassembly prediction processing on the historical service time sequence data according to the region level information to obtain a second accuracy result corresponding to a second space division granularity; determining the spatial prediction granularity from the first accuracy result and the second accuracy result.

In an exemplary embodiment of the present disclosure, performing a region aggregation prediction process on the historical service time series data according to the region level information to obtain a first accuracy result corresponding to a first space partition granularity, includes: determining a plurality of sub-areas corresponding to the target object and an area hierarchy relation among the sub-areas according to the area hierarchy information; and performing region aggregation prediction processing on the historical service time sequence data corresponding to the plurality of sub-regions according to the region hierarchical relationship to obtain a first accuracy result corresponding to the first space division granularity.

In an exemplary embodiment of the present disclosure, performing a region dismantling prediction process on the historical service time series data according to the region level information to obtain a second accuracy result corresponding to a second spatial partition granularity, includes: determining a total area and an area disassembling rule corresponding to the target object according to the area level information; and performing region dismantling prediction processing on the historical service time sequence data corresponding to the total region according to the region dismantling rule to obtain a second accuracy result corresponding to the second space division granularity.

In an exemplary embodiment of the present disclosure, determining the spatial prediction granularity from the first accuracy result and the second accuracy result comprises: acquiring an accuracy evaluation index, and determining a first prediction average value corresponding to the first accuracy result according to the accuracy evaluation index; determining a second prediction average value corresponding to the second accuracy result according to the accuracy evaluation index; comparing the first predicted average value with the second predicted average value to obtain an accuracy comparison result; and determining the spatial prediction granularity according to the accuracy comparison result.

In an exemplary embodiment of the present disclosure, the method further comprises: determining a target area, and acquiring an area constraint condition of the target area; and determining the time prediction granularity corresponding to the target area according to the area constraint condition and the accuracy comparison result.

According to a second aspect of the present disclosure, there is provided an article quantity prediction apparatus comprising: the time sequence characteristic determining module is used for acquiring historical service time sequence data of a target object and performing time sequence characteristic extraction processing on the historical service time sequence data to determine time prediction granularity corresponding to the target object; the spatial granularity determining module is used for respectively carrying out regional aggregation prediction processing and regional disassembly prediction processing on the historical service time sequence data according to the time prediction granularity so as to determine the spatial prediction granularity corresponding to the target article; and the quantity determining module is used for determining the quantity of the target item in the specific service according to the time prediction granularity and the space prediction granularity.

In an exemplary embodiment of the present disclosure, the timing characteristic determination module includes a timing characteristic determination unit for obtaining a pre-constructed timing analysis model; the time sequence analysis model is obtained based on a plurality of time sequence division data training; inputting the historical traffic timing data to the timing analysis model to determine the temporal prediction granularity from the timing analysis model; the time sequence analysis model is obtained by training the following steps: acquiring an initial time sequence model, a service time sequence training data set and a plurality of pre-configured candidate time granularities; dividing the service time sequence training data set according to the candidate time granularities to obtain a plurality of time sequence division data; and carrying out model training processing on the initial time sequence model according to the plurality of time sequence division data to obtain the time sequence analysis model.

In an exemplary embodiment of the present disclosure, the spatial granularity determination module includes a spatial granularity determination unit, configured to determine region-level information corresponding to the target item; performing region aggregation prediction processing on the historical service time sequence data according to the region level information to obtain a first accuracy result corresponding to a first space division granularity; performing region disassembly prediction processing on the historical service time sequence data according to the region level information to obtain a second accuracy result corresponding to a second space division granularity; determining the spatial prediction granularity from the first accuracy result and the second accuracy result.

In an exemplary embodiment of the present disclosure, the spatial granularity determining unit includes a first result determining subunit, configured to determine, according to the area-level information, a plurality of sub-areas corresponding to the target item and an area-level relationship between the plurality of sub-areas; and performing region aggregation prediction processing on the historical service time sequence data corresponding to the plurality of sub-regions according to the region hierarchical relationship to obtain a first accuracy result corresponding to the first space division granularity.

In an exemplary embodiment of the present disclosure, the spatial granularity determining unit includes a second result determining subunit, configured to determine, according to the region level information, a total region corresponding to the target item and a region dismantling rule; and performing region dismantling prediction processing on the historical service time sequence data corresponding to the total region according to the region dismantling rule to obtain a second accuracy result corresponding to the second space division granularity.

In an exemplary embodiment of the present disclosure, the spatial granularity determination unit includes a spatial granularity determination subunit configured to obtain an accuracy evaluation index, and determine a first prediction average corresponding to the first accuracy result according to the accuracy evaluation index; determining a second prediction average value corresponding to the second accuracy result according to the accuracy evaluation index; comparing the first prediction average value with the second prediction average value to obtain an accuracy comparison result; and determining the spatial prediction granularity according to the accuracy comparison result.

In an exemplary embodiment of the disclosure, the spatial granularity determination subunit is configured to perform: determining a target area, and acquiring an area constraint condition of the target area; and determining the time prediction granularity corresponding to the target area according to the area constraint condition and the accuracy comparison result.

According to a third aspect of the present disclosure, there is provided an electronic device comprising: a processor; and a memory having computer readable instructions stored thereon which, when executed by the processor, implement the item quantity prediction method according to any one of the above.

According to a fourth aspect of the present disclosure, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the item quantity prediction method according to any one of the above.

The technical scheme provided by the disclosure can comprise the following beneficial effects:

in the article quantity prediction method in the exemplary embodiment of the present disclosure, on one hand, the spatial prediction granularity corresponding to the target article may be determined by performing the regional aggregation prediction processing and the regional disassembly prediction processing on the historical service time series data. On the other hand, a complex scheme for determining a prediction result by combining multiple dimensions such as time prediction granularity and space prediction granularity is provided, and a scheme for predicting the specific service quantity of the target object in multiple dimensions is realized.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the disclosure. It is to be understood that the drawings in the following description are merely exemplary of the disclosure, and that other drawings may be derived from those drawings by one of ordinary skill in the art without the exercise of inventive faculty. In the drawings:

FIG. 1 schematically illustrates a flow chart of an item quantity prediction method according to an exemplary embodiment of the present disclosure;

fig. 2 schematically shows a flow chart for determining a temporal/spatial prediction scheme according to an exemplary embodiment of the present disclosure;

FIG. 3 schematically illustrates a diagram of multiple time dimension analysis of historical traffic temporal data for temporal feature extraction, according to an exemplary embodiment of the present disclosure;

FIG. 4 schematically illustrates a schematic diagram of determining a spatial prediction granularity of a target item, according to an exemplary embodiment of the present disclosure;

FIG. 5 schematically illustrates a comparison graph of average percentage error values for different regions according to an exemplary embodiment of the present disclosure;

FIG. 6 schematically illustrates a block diagram of an item quantity prediction apparatus according to an exemplary embodiment of the present disclosure;

FIG. 7 schematically illustrates a block diagram of an electronic device according to an exemplary embodiment of the present disclosure;

fig. 8 schematically illustrates a schematic diagram of a computer-readable storage medium according to an exemplary embodiment of the present disclosure.

Detailed Description

Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art. The same reference numerals denote the same or similar parts in the drawings, and thus, a repetitive description thereof will be omitted.

Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the disclosure. One skilled in the relevant art will recognize, however, that the subject matter of the present disclosure can be practiced without one or more of the specific details, or with other methods, components, devices, steps, and so forth. In other instances, well-known structures, methods, devices, implementations, materials, or operations are not shown or described in detail to avoid obscuring aspects of the disclosure.

The block diagrams shown in the figures are functional entities only and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in the form of software, or in one or more software-hardened modules, or in different networks and/or processor devices and/or microcontroller devices.

At present, most of sales prediction in the existing project is based on business scenes about time dimension and space aggregation dimension, and the prediction result of the time dimension generally comprises the time of day, week, month or quarter; for the spatial dimension, some hierarchical model schemes are about how to make the overall prediction most accurate by using Bayesian prediction. However, the sales prediction is relatively complex by adopting the combined selection of the time dimension and the space dimension, and in the existing implementation scheme, a scheme for performing the sales prediction based on the combined selection of the time dimension and the space dimension is lacked.

Based on this, in the present exemplary embodiment, first, an article quantity prediction method is provided, which may be implemented by a server, or a terminal device, where the terminal described in the present disclosure may include a mobile terminal such as a mobile phone, a tablet computer, a notebook computer, a palm computer, a Personal Digital Assistant (PDA), and a fixed terminal such as a desktop computer. Fig. 1 schematically illustrates a schematic diagram of a method flow for item quantity prediction, according to some embodiments of the present disclosure. Referring to fig. 1, the item quantity prediction method may include the steps of:

step S110, obtaining historical service time sequence data of the target object, and performing time sequence feature extraction processing on the historical service time sequence data to determine time prediction granularity corresponding to the target object;

step S120, respectively carrying out regional aggregation prediction processing and regional disassembly prediction processing on the historical service time sequence data according to the time prediction granularity so as to determine the spatial prediction granularity corresponding to the target object;

and step S130, determining the quantity of the target object in the specific service according to the time prediction granularity and the space prediction granularity.

According to the article quantity prediction method in the present example, the spatial prediction granularity corresponding to the target article can be determined by performing the regional aggregation prediction processing and the regional disassembly prediction processing on the historical service time series data. On the other hand, a complex scheme for determining a prediction result by combining multiple dimensions such as time prediction granularity and space prediction granularity is provided, and a scheme for predicting the specific service quantity of a target object in multiple dimensions is realized.

Next, the item quantity prediction method in the present exemplary embodiment will be further described.

In step S110, historical service time series data of the target item is obtained, and time series feature extraction processing is performed on the historical service time series data to determine a time prediction granularity corresponding to the target item.

In some exemplary embodiments of the present disclosure, the target item may be an item currently undergoing a specific traffic volume prediction process. The historical business time sequence data can be data generated after related business data of the target object in a specific past time period are arranged according to a time sequence. The article time sequence feature may be a related data feature obtained after feature exploration processing is performed on historical business time sequence data of the target article. The temporal prediction granularity may be a temporal granularity employed in the quantity prediction process for the particular business data of the target item. For example, the temporal prediction granularity may be one week, 10 days, 4 weeks, etc.

In order to ensure normal circulation among all links in the sales scene, the quantity of the target articles in a specific service can be determined by the article quantity prediction method disclosed by the invention, so that corresponding processing strategies are formulated for different links. In order to determine the future sales of different articles and to replenish the warehouse in time, the sales quantity of each article needs to be predicted, and after the future sales quantity of the target article is predicted, the quantity of the target article to be replenished in the warehouse can be determined in advance.

The present embodiment is illustrated by taking the determination of sales of target items as an example, and referring to fig. 2, fig. 2 schematically shows a flowchart for determining a temporal/spatial prediction scheme according to an exemplary embodiment of the present disclosure. In step S210, data is collected. Before predicting the sales volume of the target item, historical business time sequence data corresponding to the target item can be acquired. For example, sales data of the target object in a specific time period is acquired and arranged according to the time sequence to generate corresponding historical service time sequence data.

In step S220, Data exploration and Analysis (EDA). After the historical service time sequence data is obtained, the time sequence feature extraction processing can be performed on the historical service time sequence data through an EDA (electronic design automation) analysis step, so that the time prediction granularity adopted when sales prediction processing is performed on the target object is determined.

In an exemplary embodiment of the present disclosure, a pre-constructed time series analysis model is obtained; the time sequence analysis model is obtained based on a plurality of time sequence division data training; inputting historical service time sequence data into a time sequence analysis model so as to determine time prediction granularity by the time sequence analysis model; the time sequence analysis model is obtained by training the following steps: acquiring an initial time sequence model, a service time sequence training data set and a plurality of pre-configured candidate time granularities; dividing the service time sequence training data set according to the candidate time granularities to obtain a plurality of time sequence division data; and carrying out model training processing on the initial time sequence model according to the plurality of time sequence division data to obtain a time sequence analysis model.

The time sequence analysis model may be a model for performing time sequence feature analysis on the historical traffic time sequence data to determine the time prediction granularity. The time sequence division data may be data obtained by dividing a service time sequence training data set. The initial timing model may be an initially constructed timing feature extraction model. The traffic timing training data set may be a training data set used for model training of the initial timing model to obtain a timing analysis model. The candidate time granularity may be a granularity used for dividing the service timing training data set according to time length, for example, the candidate time granularity may be one day, one week, 10 days, two weeks, four weeks, or the like.

With continued reference to fig. 2, in step S201, a time series characteristic analysis is performed, i.e., the time series characteristic analysis is performed on the historical traffic time series data. For example, when the time sequence feature analysis is performed on the historical service time sequence data, the time sequence feature analysis can be performed through a time sequence analysis model which is constructed in advance. The time sequence analysis model in this embodiment performs time sequence feature extraction on the historical business time sequence data of the target object to determine the time prediction granularity corresponding to the target object. The training steps of the time sequence analysis model are as follows: when the time sequence analysis model is trained, a pre-constructed initial time sequence model and a service time sequence training data set can be obtained; the business sequence training data set can contain sales data of the target object in the historical time period. After the service time sequence training data set is determined, the pre-configured candidate time granularity can be obtained, and the service time sequence training data set is divided according to the candidate time granularity to obtain a plurality of time sequence divided data. The initial time sequence model can be trained through the time sequence division data, the time sequence characteristics in the service time sequence training data set are analyzed and learned through the model, and finally the time sequence analysis model is obtained.

Referring to fig. 3, fig. 3 schematically illustrates a diagram of multiple time dimension analysis of historical traffic temporal data for temporal feature extraction, according to an exemplary embodiment of the present disclosure. The corresponding sales amount of the target item per day is shown in fig. 3(a) with a statistical period of weeks; wherein the abscissa is the number of days corresponding to a week, and the ordinate is the sales volume of the target item. The corresponding sales volume of the target item per day for a statistical period of months is shown in FIG. 3 (b); wherein the abscissa is the number of days corresponding to one month, and the ordinate is the sales volume of the target item. FIG. 3(c) shows the sales amount of the target item per month in a statistical period of years; wherein the abscissa is the corresponding month in a year, and the ordinate is the sales volume of the target item. The corresponding sales volume of the target item per day in a statistical period of years is shown in fig. 3 (d); wherein the abscissa is the corresponding number of days in a year, and the ordinate is the sales volume of the target item. FIG. 3(e) shows the sales volume of the target item per week for a statistical period of years; wherein the abscissa is the corresponding number of weeks in a year, and the ordinate is the sales number of the target item. FIG. 3(f) shows the sales volume for each quarter of the target item in a statistical cycle of years; wherein the abscissa is the number of corresponding quarters in a year and the ordinate is the sales number of the target item.

Based on the time-series division data in fig. 3, in order to obtain an optimal time dimension unit, in this embodiment, a Pearson correlation coefficient (Pearson correlation coefficient) is used to count a correlation coefficient in each dimension, and taking the data illustrated in fig. 3 as an example, correlation coefficients of a weekly effect, a monthly/yearly effect, a daily/yearly effect, a weekly/yearly effect, and a quarterly/yearly effect are respectively obtained, as shown in table 1.

TABLE 1

Since Pearson coefficients are positive effects and the higher the effects the stronger the correlation, Pearson coefficients corresponding to weekly and monthly effects are errors that may be generated for comparing data from a particular day to a statistical period of months for the data in table 1. When the time prediction granularity is determined, comparing four correlation coefficients of a month/year effect, a day/year effect, a week/year effect, a quarter/year effect and the like, the results can be obtained, wherein the highest results are the week/year effect, the highest results are the day/year effect and the month/year effect, and the time prediction granularity is the optimal time division granularity when the sales volume of the target object is predicted.

In step S120, a regional aggregation prediction process and a regional disassembly prediction process are performed on the historical service time series data according to the temporal prediction granularity, so as to determine a spatial prediction granularity corresponding to the target article.

In some exemplary embodiments of the present disclosure, the area aggregation prediction process may be a process of performing an aggregation prediction process on time-series data from a next layer to a previous layer in an area hierarchy to determine a corresponding prediction accuracy result. The area dismantling prediction processing may be a processing procedure of carrying out dismantling prediction processing on the time series data from the previous layer to the next layer according to the area level to determine a corresponding prediction accuracy result. The spatial prediction granularity may be a spatial granularity employed in performing quantity prediction processing on specific business data of the target item. For example, the spatial prediction granularity may be configured to fit all warehouses of a city, etc.

With continued reference to fig. 2, after the temporal prediction granularity is determined, the historical business timing data may continue to be analyzed in the spatial dimension to determine a spatial prediction dimension corresponding to the target item. In step S202, the area aggregation prediction processing is performed on the historical traffic time series data. In step S203, the area dismantling prediction process is performed on the historical service time series data. After the regional aggregation prediction processing and the regional disassembly prediction processing are respectively carried out on the historical service time sequence data, the accuracy of the processing results obtained by the two processing modes is analyzed, and the spatial prediction granularity corresponding to the target object can be determined.

In an exemplary embodiment of the present disclosure, regional level information corresponding to a target item is determined; performing regional aggregation prediction processing on historical service time sequence data according to regional hierarchy information to obtain a first accuracy result corresponding to a first space division granularity; performing region disassembly prediction processing on the historical service time sequence data according to the region level information to obtain a second accuracy result corresponding to the second space division granularity; determining a spatial prediction granularity from the first accuracy result and the second accuracy result.

The region-level information may be related information of a corresponding level of a region where the warehouse for storing the articles is located, for example, the region-level information may include a division of regions of different levels. The first space division granularity may be determined by performing region aggregation prediction processing on historical service time series data. The first accuracy result may be an accuracy result corresponding to when the quantity of the item is predicted according to the first spatial division granularity. The second spatial partition granularity may be determined by performing region decomposition prediction processing on the historical service time series data. The second accuracy result may be a corresponding accuracy result when the quantity of the item is predicted according to the second spatial division granularity.

Referring to fig. 4, fig. 4 schematically illustrates a schematic diagram of determining a spatial prediction granularity of a target item, according to an exemplary embodiment of the present disclosure. In step S410, the region level information corresponding to the target item is determined. When determining the spatial prediction granularity of the target item, the regional level information corresponding to the target item may be determined, where the regional level information may include distribution of warehouses storing the items in different regions and a hierarchical relationship between warehouses in different regions. With continued reference to fig. 2, before performing the area aggregation prediction process and the area disassembly prediction process on the historical service time series data, in step S230, a feature engineering process is performed. Specifically, the historical service time sequence data is subjected to relevant processing such as feature extraction, feature selection, feature construction and the like. For example, a warehouse of different areas corresponding to the target item is characterized.

In step S420, a region aggregation prediction process is performed on the historical service time series data according to the region level information, so as to obtain a first accuracy result corresponding to the first space division granularity. The regional aggregation prediction processing is performed based on the characterized historical business time series data, and specifically, the regional aggregation prediction processing (bottom up, BU) may be a processing mode in which sales data related to the target item are aggregated from a lower layer to an upper layer, for example, sales data of each region are aggregated nationwide. A first accuracy result corresponding to the target item determined based on the first space division granularity may be determined by the region aggregation prediction process.

In step S430, a region decomposition prediction process is performed on the historical service time series data according to the region level information, so as to obtain a second accuracy result corresponding to the second space partition granularity. The area dismantling prediction process (TD) may be a process of dismantling data related to a target object from upper layers to lower layers, for example, a process of dismantling national sales data into areas. A second accuracy result corresponding to the target item determined based on the second spatial granularity may be determined by the region deconstruction prediction process.

In step S440, a spatial prediction granularity is determined according to the first accuracy result and the second accuracy result. By comparing the first accuracy result with the second accuracy result, a spatial prediction granularity may be determined from the accuracy comparison result.

In an exemplary embodiment of the present disclosure, a region hierarchical relationship between a plurality of sub-regions and a plurality of sub-regions corresponding to a target article is determined according to region hierarchical information; and performing region aggregation prediction processing on historical service time sequence data corresponding to the plurality of sub-regions according to the region hierarchical relation to obtain a first accuracy result corresponding to the first space division granularity.

The sub-regions may be regions of each region formed by dividing the warehouse according to the region level information. The regional hierarchical relationship may be a hierarchical relationship of sub-regions between respective levels, for example, the regional hierarchical relationship may include a hierarchical relationship in which respective article depositories divide regions based on administrative levels of nationwide, provincial, municipal, regional, and the like.

After the region level information is acquired, the sub-regions corresponding to the target articles, that is, the warehouses of the province and city regions where the warehouses storing the target articles are located, may be determined according to the region level information, and the region level relationship between the sub-regions is determined. For example, for a target object, due to the fact that stores selling the target object may be included in different streets, store sales data of all the sold target objects in one street may be summarized, and after the store sales data of all the streets in one area are summarized, sales corresponding to one area may be obtained. At this time, the sales data of the plurality of areas may be continuously summarized to obtain the sales amount corresponding to one city. In this embodiment, the area hierarchical relationship between the multiple sub-areas may be determined according to the regional administrative relationship, the sales of the total area may be predicted according to the area hierarchical relationship and the sales data corresponding to the multiple sub-areas, and the obtained predicted value may be compared with the actual value corresponding to the total area to determine the first accuracy result.

In an exemplary embodiment of the disclosure, a total area corresponding to a target article and an area dismantling rule are determined according to area level information; and performing region dismantling prediction processing on the historical service time sequence data corresponding to the total region according to a region dismantling rule to obtain a second accuracy result corresponding to the second space division granularity.

The total area may be an area corresponding to the storage of all sub-areas of the storage object. The region splitting rule may be a rule adopted for splitting the total region to form a plurality of sub-regions of different levels.

When performing region dismantling prediction processing on historical service time series data, a total region corresponding to a target article can be determined according to region level information, for example, the total region is generally taken nationwide as the target article; and acquiring a region dismantling rule for the total region from the region level information, for example, the nationwide region has a total sales data, and according to the region characteristics of different sub-regions, corresponding sales shares can be allocated to the sub-regions, thereby determining the sales data of all sub-regions included in the total region. According to the data obtained through the processing, prediction processing can be performed on the sales data corresponding to each sub-region to obtain a predicted value corresponding to each sub-region, the obtained predicted value is compared with an actual value, and a second accuracy result obtained after the region is divided by adopting a second space division granularity is determined.

In an exemplary embodiment of the present disclosure, an accuracy evaluation index is obtained, and a first predicted average corresponding to the first accuracy result is determined according to the accuracy evaluation index; determining a second prediction average value corresponding to a second accuracy result according to the accuracy evaluation index; comparing the first prediction average value with the second prediction average value to obtain an accuracy comparison result; and determining the spatial prediction granularity according to the accuracy comparison result.

The accuracy evaluation index may be an index used for evaluating an accuracy result. The first predicted average may be a result of averaging the first accuracy results. The second predicted average may be a result of averaging the second accuracy results. The accuracy comparison result may be a comparison of the first predicted average value and the second predicted average value.

With continued reference to FIG. 2, at step S240, a multi-model training operation is performed. In the present disclosure, model training may be performed by a region aggregation prediction process, a region disassembly prediction process, and the like, so as to determine a model for determining a spatial prediction granularity. In step S204, a spatial test result comparison is performed. Specifically, the first accuracy result may be compared with the second accuracy result, and the spatial prediction granularity may be determined according to the comparison result. Specifically, before the accuracy comparison, the accuracy evaluation index may be determined, and in this embodiment, a Mean Absolute Percentage Error (MAPE) may be used as the accuracy evaluation index. Referring to table 2, the accuracy prediction results of each region with week as the statistical period are shown in table 2; the first prediction average value is included, namely data corresponding to the area average value column; and a second predicted average, i.e., data corresponding to a national column.

TABLE 2

Line label	Area mean (TD)	China (BU)	In total
				0	48％	22％	46％
1	31％	32％	31％
				2	38％	34％	36％
3	32％	34％	33％
				4	34％	35％	34％
5	34％	41％	38％
				6	36％	42％	39％
7	32％	45％	39％
				8	34％	64％	49％
9	33％	65％	49％
				10	35％	58％	46％
11	35％	63％	49％
				12	34％	75％	54％
13	35％	77％	56％
				14	35％	100％	66％
15	34％	118％	76％
				16	34％	131％	83％
17	33％	138％	87％
				18	37％	147％	92％
19	34％	149％	91％
				20	33％	152％	94％
Total of	34％	77％	55％

As can be seen from the analysis of the data in Table 2, the data in Table 2 is the map value statistics of region aggregation (bottom up) and region disassembly (top down) in 21 weeks of testing (data lines from 0-20); wherein 34% of the result of region aggregation (bottom up) obtained in the total row is better than 77% of the result of region disassembly (top down), and the accuracy rate is generally optimal when the periodic row label is 1, namely, the 2 nd period, wherein the region is 31% and the country is 32%. Therefore, the spatial prediction granularity determined according to the regional aggregation prediction processing mode has high accuracy, and the spatial prediction granularity can be determined by adopting the regional aggregation processing mode.

In an exemplary embodiment of the present disclosure, a target area is determined, and an area constraint condition of the target area is obtained; and determining the time prediction granularity corresponding to the target region according to the region constraint condition and the accuracy comparison result.

The target area may be an area where an item quantity prediction is to be performed. The regional constraint can be a decision input condition for the target region to predict the quantity of the item, for example, the regional constraint can be a constraint determined by the supply chain for the target region.

With continued reference to FIG. 2, in step S250, a model selection operation. Specifically, in step S205, time test result comparison is performed, for example, an optimal model is determined according to output results of a plurality of models. Referring to fig. 5, fig. 5 schematically shows a comparison of the average percent error values for different regions according to an exemplary embodiment of the present disclosure. As can be seen from fig. 5, the regions 0-2 and 5 are stable in performance at different periods, and the

regions

3, 4 and 6 have large fluctuation, but the map value is small at the period 1, i.e., the second period, so that the overall consideration is that 2 periods can be uniformly selected as the temporal prediction granularity. In step S206, a prediction scheme is output. For example, in this embodiment, the spatial prediction granularity is determined according to a spatial aggregation processing manner, and the temporal prediction granularity corresponding to all the regions is determined as a week.

Further, due to the specificity of the partial regions, in some scenarios, the corresponding temporal prediction granularity may be determined separately for some regions. With continuing reference to fig. 5, it can be seen from fig. 5 that region No. 4 is at greater risk and the map value is increased for the third week, so for region No. 4, 4 weeks can be selected as the prediction period, i.e., the temporal prediction granularity is determined to be 4 weeks.

In step S130, the quantity of the target item in the specific service is determined according to the temporal prediction granularity and the spatial prediction granularity.

In some exemplary embodiments of the present disclosure, the specific business may be a specific business corresponding to the target item, for example, the specific business may be a future sales amount of the target item, the specific business may also be a warehouse to-be-restocked amount of the target item, and the like.

After the time prediction granularity and the space prediction granularity are determined, the quantity prediction can be performed on the specific business of the target object according to the determined time prediction granularity and the determined space prediction granularity so as to determine the quantity of the target object in the specific business. With continued reference to fig. 2, in step S260, the result is output. After the quantity of the target object in the specific service is determined according to the time prediction granularity and the space prediction granularity, the corresponding quantity of the object can be output.

It should be noted that the terms "first", "second", "third", "fourth", etc. used in this disclosure are only used for distinguishing different accuracy results, different spatial partition granularity, different prediction average values, and should not impose any limitation on this disclosure.

In summary, the article quantity prediction method disclosed by the present disclosure obtains historical service time sequence data of the target article, and performs time sequence feature extraction processing on the historical service time sequence data to determine a time prediction granularity corresponding to the target article; respectively carrying out regional aggregation prediction processing and regional disassembly prediction processing on the historical service time sequence data according to the time prediction granularity so as to determine the spatial prediction granularity corresponding to the target object; and determining the quantity of the target object in the specific business according to the time prediction granularity and the space prediction granularity. On one hand, the spatial prediction granularity corresponding to the target object can be determined by performing the regional aggregation prediction processing and the regional disassembly prediction processing on the historical service time sequence data. On the other hand, a complex scheme for determining a prediction result by combining multiple dimensions such as time prediction granularity and space prediction granularity is provided, and a scheme for predicting the specific service quantity of the target object in multiple dimensions is realized. In yet another aspect, by providing sku-level-accurate configuration, refinement of energized product operation can lead to predicted results under superior temporal and spatial configuration schemes.

It is noted that although the steps of the methods of the present invention are depicted in the drawings in a particular order, this does not require or imply that the steps must be performed in this particular order, or that all of the depicted steps must be performed, to achieve desirable results. Additionally or alternatively, certain steps may be omitted, multiple steps combined into one step execution, and/or one step broken into multiple step executions, etc.

Further, in the present exemplary embodiment, an article quantity prediction apparatus is also provided. Referring to fig. 6, the item quantity prediction apparatus 600 may include: a timing characteristic determination module 610, a spatial granularity determination module 620, and a quantity determination module 630.

Specifically, the time sequence feature determining module 610 is configured to obtain historical service time sequence data of the target object, and perform time sequence feature extraction processing on the historical service time sequence data to determine a time prediction granularity corresponding to the target object; the spatial granularity determining module 620 is configured to perform region aggregation prediction processing and region disassembly prediction processing on the historical service time sequence data according to the time prediction granularity to determine a spatial prediction granularity corresponding to the target article; the quantity determining module 630 is configured to determine the quantity of the target item in the specific service according to the temporal prediction granularity and the spatial prediction granularity.

In an exemplary embodiment of the present disclosure, the timing characteristic determination module includes a timing characteristic determination unit for obtaining a pre-constructed timing analysis model; the time sequence analysis model is obtained based on a plurality of time sequence division data training; inputting historical service time sequence data into a time sequence analysis model so as to determine time prediction granularity by the time sequence analysis model; the time sequence analysis model is obtained by training the following steps: acquiring an initial time sequence model, a service time sequence training data set and a plurality of pre-configured candidate time granularities; dividing the service time sequence training data set according to the candidate time granularities to obtain a plurality of time sequence division data; and performing model training processing on the initial time sequence model according to the plurality of time sequence division data to obtain a time sequence analysis model.

In an exemplary embodiment of the present disclosure, the spatial granularity determination module includes a spatial granularity determination unit, configured to determine region-level information corresponding to the target item; performing regional aggregation prediction processing on historical service time sequence data according to regional hierarchy information to obtain a first accuracy result corresponding to a first space division granularity; performing region disassembly prediction processing on the historical service time sequence data according to the region level information to obtain a second accuracy result corresponding to the second space division granularity; determining a spatial prediction granularity from the first accuracy result and the second accuracy result.

In an exemplary embodiment of the present disclosure, the spatial granularity determining unit includes a first result determining subunit, configured to determine, according to the area-level information, an area-level relationship between a plurality of sub-areas corresponding to the target item and the plurality of sub-areas; and performing region aggregation prediction processing on historical service time sequence data corresponding to the plurality of sub-regions according to the region hierarchical relation to obtain a first accuracy result corresponding to the first space division granularity.

In an exemplary embodiment of the present disclosure, the spatial granularity determining unit includes a second result determining subunit, configured to determine, according to the region level information, a total region corresponding to the target item and a region dismantling rule; and performing region dismantling prediction processing on the historical service time sequence data corresponding to the total region according to a region dismantling rule to obtain a second accuracy result corresponding to the second space division granularity.

In an exemplary embodiment of the disclosure, the spatial granularity determination subunit is configured to perform: determining a target area, and acquiring an area constraint condition of the target area; and determining the time prediction granularity corresponding to the target region according to the region constraint condition and the accuracy comparison result.

The details of the virtual module of each item quantity prediction apparatus are already described in detail in the corresponding item quantity prediction method, and therefore are not described herein again.

It should be noted that although several modules or units of the article quantity predicting device are mentioned in the above detailed description, such division is not mandatory. Indeed, the features and functionality of two or more modules or units described above may be embodied in one module or unit, according to embodiments of the present disclosure. Conversely, the features and functions of one module or unit described above may be further divided into embodiments by a plurality of modules or units.

In addition, in an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above method is also provided.

As will be appreciated by one skilled in the art, aspects of the present invention may be embodied as a system, method or program product. Thus, various aspects of the invention may be embodied in the form of: an entirely hardware embodiment, an entirely software embodiment (including firmware, microcode, etc.) or an embodiment combining hardware and software aspects that may all generally be referred to herein as a "circuit," module "or" system.

An electronic device 700 according to such an embodiment of the disclosure is described below with reference to fig. 7. The electronic device 700 shown in fig. 7 is only an example and should not bring any limitation to the functions and the scope of use of the embodiments of the present disclosure.

As shown in fig. 7, electronic device 700 is embodied in the form of a general purpose computing device. The components of the electronic device 700 may include, but are not limited to: the at least one processing unit 710, the at least one memory unit 720, a bus 730 connecting different system components (including the memory unit 720 and the processing unit 710), and a display unit 740.

Wherein the storage unit stores program code that is executable by the processing unit 710 to cause the processing unit 710 to perform steps according to various exemplary embodiments of the present disclosure as described in the above section "exemplary methods" of this specification.

The storage unit 720 may include readable media in the form of volatile memory units, such as a random access memory unit (RAM)721 and/or a cache memory unit 722, and may further include a read only memory unit (ROM) 723.

The memory unit 720 may include a program/utility 724 having a set (at least one) of program modules 725, such program modules 725 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination thereof may comprise an implementation of a network environment.

Bus 730 may represent one or more of any of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

The electronic device 700 may also communicate with one or more external devices 770 (e.g., keyboard, pointing device, bluetooth device, etc.), with one or more devices that enable a user to interact with the electronic device 700, and/or with any devices (e.g., router, modem, etc.) that enable the electronic device 700 to communicate with one or more other computing devices. Such communication may occur via an input/output (I/O) interface 750. Also, the electronic device 700 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network such as the internet) via the network adapter 760. As shown, the network adapter 760 communicates with the other modules of the electronic device 700 via the bus 730. It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.

Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solution according to the embodiments of the present disclosure may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (which may be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which may be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiments of the present disclosure.

In an exemplary embodiment of the present disclosure, there is also provided a computer-readable storage medium having stored thereon a program product capable of implementing the above-described method of the present specification. In some possible embodiments, aspects of the invention may also be implemented in the form of a program product comprising program code means for causing a terminal device to carry out the steps according to various exemplary embodiments of the invention described in the above-mentioned "exemplary methods" section of the present description, when said program product is run on the terminal device.

Referring to fig. 8, a program product 800 for implementing the above method according to an embodiment of the present invention is described, which may employ a portable compact disc read only memory (CD-ROM) and include program code, and may be run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited in this regard and, in the present document, a readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

A computer readable signal medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable signal medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

Program code embodied on a readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).

Furthermore, the above-described figures are merely schematic illustrations of processes involved in methods according to exemplary embodiments of the invention, and are not intended to be limiting. It will be readily understood that the processes shown in the above figures are not intended to indicate or limit the chronological order of the processes. In addition, it is also readily understood that these processes may be performed synchronously or asynchronously, e.g., in multiple modules.

Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

It will be understood that the present disclosure is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is to be limited only by the terms of the appended claims.

Claims

1. A method for predicting quantity of an item, comprising:

acquiring historical service time sequence data of a target article, and performing time sequence feature extraction processing on the historical service time sequence data to determine a time prediction granularity corresponding to the target article;

respectively carrying out regional aggregation prediction processing and regional disassembly prediction processing on the historical service time sequence data according to the time prediction granularity so as to determine the spatial prediction granularity corresponding to the target article;

and determining the item quantity of the target item in a specific service according to the temporal prediction granularity and the spatial prediction granularity.

2. The method according to claim 1, wherein performing a time series feature extraction process on the historical traffic time series data to determine a time prediction granularity corresponding to the target item comprises:

acquiring a pre-constructed time sequence analysis model; the time sequence analysis model is obtained based on a plurality of time sequence division data training;

inputting the historical traffic timing data to the timing analysis model to determine the temporal prediction granularity from the timing analysis model;

the time sequence analysis model is obtained by training the following steps:

acquiring an initial time sequence model, a service time sequence training data set and a plurality of pre-configured candidate time granularities;

dividing the service time sequence training data set according to the candidate time granularities to obtain a plurality of time sequence division data;

and carrying out model training processing on the initial time sequence model according to the plurality of time sequence division data to obtain the time sequence analysis model.

3. The method according to claim 1, wherein performing a regional aggregation prediction process and a regional disassembly prediction process on the historical service time series data according to the temporal prediction granularity to determine a spatial prediction granularity corresponding to the target item comprises:

determining the region level information corresponding to the target object;

performing region aggregation prediction processing on the historical service time sequence data according to the region level information to obtain a first accuracy result corresponding to a first space division granularity;

performing region disassembly prediction processing on the historical service time sequence data according to the region level information to obtain a second accuracy result corresponding to a second space division granularity;

determining the spatial prediction granularity from the first accuracy result and the second accuracy result.

4. The method according to claim 3, wherein performing a region aggregation prediction process on the historical traffic time series data according to the region level information to obtain a first accuracy result corresponding to a first spatial partition granularity, comprises:

determining a plurality of sub-areas corresponding to the target object and an area hierarchy relation among the sub-areas according to the area hierarchy information;

and performing region aggregation prediction processing on the historical service time sequence data corresponding to the plurality of sub-regions according to the region hierarchical relationship to obtain a first accuracy result corresponding to the first space division granularity.

5. The method of claim 3, wherein performing region decomposition prediction processing on the historical traffic time series data according to the region level information to obtain a second accuracy result corresponding to a second spatial partition granularity, comprises:

determining a total area and an area disassembling rule corresponding to the target object according to the area level information;

and performing region dismantling prediction processing on the historical service time sequence data corresponding to the total region according to the region dismantling rule to obtain a second accuracy result corresponding to the second space division granularity.

6. The method of claim 3, wherein determining the spatial prediction granularity from the first accuracy result and the second accuracy result comprises:

acquiring an accuracy evaluation index, and determining a first prediction average value corresponding to the first accuracy result according to the accuracy evaluation index;

determining a second prediction average value corresponding to the second accuracy result according to the accuracy evaluation index;

comparing the first prediction average value with the second prediction average value to obtain an accuracy comparison result;

and determining the spatial prediction granularity according to the accuracy comparison result.

7. The method of claim 6, further comprising:

determining a target area, and acquiring an area constraint condition of the target area;

and determining the time prediction granularity corresponding to the target area according to the area constraint condition and the accuracy comparison result.

8. An article quantity prediction apparatus, comprising:

the time sequence characteristic determining module is used for acquiring historical service time sequence data of a target article and extracting time sequence characteristics of the historical service time sequence data to determine time prediction granularity corresponding to the target article;

the spatial granularity determining module is used for respectively carrying out regional aggregation prediction processing and regional disassembly prediction processing on the historical service time sequence data according to the time prediction granularity so as to determine the spatial prediction granularity corresponding to the target article;

and the quantity determining module is used for determining the quantity of the target item in the specific service according to the time prediction granularity and the space prediction granularity.

9. An electronic device, comprising:

a processor; and

a memory having computer readable instructions stored thereon which, when executed by the processor, implement the item quantity prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the item quantity prediction method according to any one of claims 1 to 7.