Sales information prediction method and device

By obtaining the scene attribute information and feature information of new products, distinguishing different prediction scenarios, and building differentiated models to predict new products sales volumes, it solves the problem of insufficient accuracy and interpretability of new products sales volume prediction in the existing technology, and achieves more accurate and capable The prediction results of the letter.

CN118967213APending Publication Date: 2024-11-15SHANGHAI 100 METERS NETWORK TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411181214.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The existing new product sales forecasting plan fails to make full use of new product characteristic information, resulting in insufficient accuracy and interpretability of the prediction results, especially when predicting new product sales in different scenarios.

Method used

By obtaining the scene attribute information of the new product to be tested, determining its target prediction scenario, and building corresponding new product prediction models based on different scenarios, using the feature information of candidate products and stores for differentiated modeling to predict the target sales of new products.

Benefits of technology

It improves the accuracy and interpretability of the new product sales forecast model, makes full use of the attribute information of new products and related stores, and enhances the objectivity and interpretability of the prediction results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118967213A_ABST
    Figure CN118967213A_ABST
Patent Text Reader

Abstract

The embodiment of the invention provides a sales volume information prediction method and device, which can be applied to the technical field of artificial intelligence, in the method, scene attribute information associated with a new product prediction scene and corresponding to a to-be-tested new product is acquired, and a sales volume prediction result is obtained based on the scene attribute information corresponding to the to-be-tested new product. And determining a target prediction scene corresponding to the to-be-tested new product from the plurality of new product prediction scenes, different new product prediction scenes corresponding to different new product prediction models, and predicting target sales information of the to-be-tested new product by using the new product prediction model corresponding to the target prediction scene of the to-be-tested new product. Therefore, according to different new product prediction scenes corresponding to different new product prediction models, the target sales information of the to-be-tested new product is predicted, the feature information of different new products is fully utilized, the information loss is reduced, and the interpretability of the new product prediction model is further ensured while the accuracy of the new product prediction model is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a sales information prediction method and device. Background Art

[0002] With the rapid development of artificial intelligence technology, it has a profound impact on all walks of life, especially the field of Internet technology. In the existing technology, new product prediction solutions generally use products with sales (old products) to infer the sales of unsold products (new products). For example, using the inherent attributes of the product such as origin, brand, category, price and other information as features, sales as the learning target, and using products with sales records to build a regression model. Then, the same set of features of the new product is input into the regression model to predict the sales of the new product.

[0003] However, in actual applications, there is a need to predict the sales volume of new products in different scenarios. When using the above fixed new product prediction scheme to predict the sales volume of new products in different scenarios, the feature information of new products is not fully utilized. Failure to fully utilize the feature information of new products is likely to cause information loss during modeling and prediction, and affect the accuracy of the prediction results. Summary of the invention

[0004] The embodiments of the present application provide a sales information prediction method and device to improve the accuracy of a new product sales prediction model while improving the interpretability of the new product sales prediction model.

[0005] On the one hand, an embodiment of the present application provides a sales information prediction method, the method comprising:

[0006] Obtaining scene attribute information associated with the new product prediction scene corresponding to the new product to be tested;

[0007] Based on the scene attribute information corresponding to the new product to be tested, determining a target prediction scene corresponding to the new product to be tested from multiple new product prediction scenes, where different new product prediction scenes correspond to different new product prediction models;

[0008] The new product prediction model corresponding to the target prediction scenario is used to predict the target sales information of the new product to be tested.

[0009] On the one hand, an embodiment of the present application provides a sales information prediction device, the device comprising:

[0010] A receiving module, used to obtain scene attribute information associated with a new product prediction scene corresponding to the new product to be tested;

[0011] The judgment module is used to determine the target prediction scene corresponding to the new product to be tested from multiple new product prediction scenes based on the scene attribute information corresponding to the new product to be tested, and different new product prediction scenes correspond to different new product prediction models.

[0012] The processing module is used to use the new product prediction model corresponding to the target prediction scenario to predict the target sales information of the new product to be tested.

[0013] Optionally, the scene attribute information includes: listing information;

[0014] The judgment module is specifically used for:

[0015] If the listing information indicates that the new product to be tested has been listed, a new product prediction scenario with sales volume among the multiple new product prediction scenarios is used as a target prediction scenario corresponding to the new product to be tested.

[0016] Optionally, the scene attribute information also includes: historical sales information;

[0017] The judging module is also used for:

[0018] If the listing information indicates that the new product to be tested has not been listed, and the historical sales information indicates that the new product to be tested has historical sales, then the seasonal new product prediction scenario among the multiple new product prediction scenarios is used as the target prediction scenario corresponding to the new product to be tested.

[0019] Optionally, the scene attribute information further includes: reference sales information;

[0020] The judging module is also used for:

[0021] If the listing information indicates that the new product to be tested has not been listed, the historical sales information indicates that the new product to be tested has no historical sales, and the reference sales information indicates that the new product to be tested has reference sales, then the reference new product prediction scenario among the multiple new product prediction scenarios is used as the target prediction scenario corresponding to the new product to be tested;

[0022] If the listing information indicates that the new product to be tested has not been listed, the historical sales information indicates that the new product to be tested has no historical sales, and the reference sales information indicates that the new product to be tested has no reference sales, then the pure new product prediction scenario among the multiple new product prediction scenarios is used as the target prediction scenario corresponding to the new product to be tested.

[0023] Optionally, the target prediction scenario is a pure new product prediction scenario; and the processing module is specifically used to:

[0024] Selecting a reference commodity from the plurality of candidate commodities based on the similarity between the first feature information of the new commodity to be tested and the first feature information of a plurality of candidate commodities;

[0025] Based on the historical sales information and price elasticity coefficient of the reference product, the target sales information of the new product to be tested is predicted.

[0026] Optionally, the target prediction scenario is a seasonal new product prediction scenario;

[0027] The processing module is specifically used for:

[0028] Acquire second feature information of multiple candidate commodities within a preset time range;

[0029] Selecting a reference commodity from the plurality of candidate commodities based on the similarity between the second feature information of the new commodity to be tested and the second feature information of the plurality of candidate commodities;

[0030] Determining a first adjustment coefficient based on the discount information of the reference product and the discount information of the new product to be tested;

[0031] Based on the historical sales information of the reference product and the first adjustment coefficient, the target sales information of the new product to be tested is predicted.

[0032] Optionally, the target prediction scenario is an alternative new product prediction scenario;

[0033] The processing module is specifically used for:

[0034] Obtaining historical sales information of reference products that have a substitution relationship with the new product to be tested;

[0035] Based on the historical sales information of the reference product and the preset adjustment coefficient, the target sales information of the new product to be tested is predicted.

[0036] Optionally, the target prediction scenario is a new product prediction scenario with sales volume;

[0037] The processing module is specifically used for:

[0038] If the shelf time of the new product to be tested is less than the first preset time, determining a second adjustment coefficient based on historical sales information of reference products of the same type as the new product to be tested;

[0039] Predicting target sales information of the new product to be tested based on the sales growth coefficient of the new product to be tested within the first preset time period, the historical sales information of the new product to be tested, and the second adjustment coefficient;

[0040] If the shelf time of the new product to be tested is longer than the second preset time, a machine learning regression model is used to predict the target sales information of the new product to be tested based on the third feature information of the new product to be tested and the historical sales information of the new product to be tested, and the second preset time is longer than the first preset time.

[0041] Optionally, the target prediction scenario is the reference new product prediction scenario;

[0042] The processing module is specifically used for:

[0043] Based on the similarity between the attribute information of the target store associated with the new product to be tested and the attribute information of multiple candidate stores, a reference store is selected from the multiple candidate stores;

[0044] Determining a third adjustment coefficient based on difference information between the reference store and the target store;

[0045] Based on the historical sales information of the reference store associated with the new product to be tested and the third adjustment coefficient, the target sales information of the new product to be tested is predicted.

[0046] In one aspect, an embodiment of the present application provides a computer device, including:

[0047] Memory for storing computer programs;

[0048] The processor is used to call the computer program stored in the memory and execute the steps of the above-mentioned sales information prediction method according to the obtained program.

[0049] On the one hand, an embodiment of the present application provides a computer-readable storage medium, which stores a program. When the program is run on a computer, the computer implements the steps of executing the above-mentioned sales information prediction method.

[0050] In an embodiment of the present application, new product prediction scenarios are divided according to different types of new products, which not only effectively utilizes the characteristic information of the new product to be tested itself, but also makes full use of the attribute information of the target store associated with the new product to be tested and the characteristic information of the reference product, thereby improving the accuracy of the target sales information predicted for the new product to be tested in this scenario, and at the same time improving the performance of the new product sales prediction model, thereby confirming the objectivity and interpretability of the new product sales prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0052] Figure 1 A system architecture diagram provided for an embodiment of the present invention;

[0053] Figure 2 A flow chart of a sales volume information prediction process provided by an embodiment of the present invention;

[0054] Figure 3A schematic diagram of a process for determining a commodity prediction scenario provided by an embodiment of the present invention;

[0055] Figure 4 A schematic diagram of enumerating third characteristic information provided by an embodiment of the present invention;

[0056] Figure 5 A schematic diagram of regression model training and prediction provided by an embodiment of the present invention;

[0057] Figure 6 A schematic diagram of a before-and-after comparison of a store merchandise copy provided by an embodiment of the present invention;

[0058] Figure 7 A schematic diagram of the structure of a sales information prediction device provided by an embodiment of the present invention;

[0059] Figure 8 A schematic diagram of the structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] The following is a brief introduction to the system architecture diagram applicable to the technical solution of the embodiment of the present application. It should be noted that the process introduced below is only used to illustrate the embodiment of the present application and is not a limitation.

[0062] refer to Figure 1 , which is a system architecture diagram applicable to an embodiment of the present application. The system architecture includes at least a terminal device 101 and a server 102. The number of terminal devices 101 can be one or more, and the number of servers 102 can also be one or more. The present application does not specifically limit the number of terminal devices 101 and servers 102.

[0063] The terminal device 101 can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart home appliance, an intelligent voice interaction device, an intelligent vehicle-mounted device, etc., but is not limited thereto.

[0064] Server 102 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), as well as big data and artificial intelligence platforms, but is not limited to these.

[0065] It should be noted that the method in the embodiment of the present application can be executed by the terminal device 101 or the server 102 alone, or can be executed by the terminal device 101 and the server 102 together.

[0066] In the embodiment of the present application, the terminal device 101 and the server 102 may be directly or indirectly connected to each other through one or more networks. The network may be a wired network or a wireless network, for example, a mobile cellular network, or a wireless fidelity (Wireless-Fidelity, WIFI) network, or other possible networks, which are not limited in the embodiment of the present application.

[0067] based on Figure 1 The system architecture diagram shown in the figure, the embodiment of the present application provides a process of sales information prediction, such as Figure 2 As shown, the process of the method is executed by a computer device, which may be Figure 1 The terminal device 101 and / or server 102 shown includes the following steps:

[0068] Step 201, obtaining scene attribute information associated with a new product prediction scene corresponding to the new product to be tested.

[0069] Specifically, the scene attribute information includes: shelf information, historical sales information, reference sales information, etc.; among them, the shelf information is used to indicate whether the new product to be tested has been put on the shelves, the historical sales information is used to indicate whether the new product to be tested has historical sales within two years, and the reference sales information is used to indicate whether the new product to be tested has been put on the shelves in other stores or has not been put on the shelves in other stores but has historical sales.

[0070] Step 202, based on the scene attribute information corresponding to the new product to be tested, determine the target prediction scene corresponding to the new product to be tested from multiple new product prediction scenes, and different new product prediction scenes correspond to different new product prediction models.

[0071] Specifically, multiple new product prediction scenarios include: pure new product prediction scenario, seasonal new product prediction scenario, alternative new product prediction scenario, new product prediction scenario with sales, new product prediction scenario with reference, alternative new product prediction scenario, etc. The pure new product prediction scenario is a prediction scenario for new products that are about to be put on the shelves and have no historical sales; the seasonal new product prediction scenario is a prediction scenario for new products that have not been put on the shelves recently but have been put on the shelves before, and are new products that need to be put on the shelves again after being taken off the shelves for a long time; the alternative new product prediction scenario is a prediction scenario for new products that are manually specified to be substitutes for another product; the new product prediction scenario with sales is a prediction scenario for new products that have been on the shelves for a period of time; the new product prediction scenario with reference is a prediction scenario for new products that have never been put on the shelves in the associated target stores, but have historical sales information in other reference stores.

[0072] In some embodiments, the present application uses at least the following routing method to determine the target prediction scenario corresponding to the new product to be tested, such as Figure 3 As shown, the following steps are included:

[0073] Step 301, obtaining the shelf information, historical sales information, and reference sales information corresponding to the new product to be tested.

[0074] Step 302 , determining whether the shelf information indicates that the new product to be tested has been put on the shelf, if so, executing step 303 , otherwise executing step 304 .

[0075] For example, when the listing information shows that the new product to be tested has historical sales in the past seven days, it means that the new product to be tested has been listed. When the listing information shows that the new product to be tested has no historical sales in the past seven days, it means that the new product to be tested has not been listed.

[0076] Step 303: Select a new product prediction scenario with sales volume from multiple new product prediction scenarios as a target prediction scenario corresponding to the new product to be tested.

[0077] Step 304 , determining whether the historical sales information indicates that the new product to be tested has historical sales, if so, executing step 305 , otherwise executing step 306 .

[0078] For example, when the historical sales information shows that the new product to be tested has a sales record within two years, it means that the new product to be tested has historical sales. When the historical sales information shows that the new product to be tested has no sales record within two years, it means that the new product to be tested has no historical sales.

[0079] Step 305 , selecting a seasonal new product prediction scenario from a plurality of new product prediction scenarios as a target prediction scenario corresponding to the new product to be tested.

[0080] Step 306 , determining whether the reference sales information indicates that the new product to be tested has reference sales. If so, executing step 307 , otherwise executing step 308 .

[0081] For example, when the reference sales information shows that the new product to be tested has sales records in other stores, it means that the new product to be tested has reference sales. When the reference sales information shows that the new product to be tested has no sales records in other stores, it means that the new product to be tested has no reference sales.

[0082] Step 307 : selecting a reference new product prediction scenario from a plurality of new product prediction scenarios as a target prediction scenario corresponding to the new product to be tested.

[0083] Step 308 , selecting a pure new product prediction scenario from the multiple new product prediction scenarios as a target prediction scenario corresponding to the new product to be tested.

[0084] In the embodiment of the present application, different prediction scenarios for the new products to be tested are divided, especially to distinguish pure new products from other types of new products, thereby reducing the loss of characteristic information of the new products to be tested themselves, providing a basis for subsequent differentiated modeling for different prediction scenarios, and further enhancing the interpretability of the sales prediction model for the new products to be tested.

[0085] Step 203: Use the new product prediction model corresponding to the target prediction scenario to predict the target sales information of the new product to be tested.

[0086] In the embodiment of the present application, the target prediction scenario corresponding to the new product to be tested is determined by the scenario attribute information, the different branches of the new product to be tested are distinguished, the information of the new product to be tested itself is fully utilized, and the probability of inaccurate target prediction sales prediction of the new product to be tested due to information loss is reduced. According to the different target prediction scenarios corresponding to the new product to be tested, differentiated modeling is used to predict the target prediction sales information of the new product to be tested, instead of using the same set of feature information and a single regression prediction model. Compared with the prior art, this differentiated modeling for different target prediction scenarios solves the interpretability problem of the prediction model, thereby improving the accuracy of the sales information prediction of the new product to be tested.

[0087] In some embodiments, when the target prediction scenario is a pure new product prediction scenario, the present application uses the following method to predict the target sales information of the new product to be tested:

[0088] Based on the similarity between the first feature information of the new product to be tested and the first feature information of multiple candidate products, a reference product is selected from the multiple candidate products. Based on the historical sales information and price elasticity coefficient of the reference product, the target sales information of the new product to be tested is predicted.

[0089] Specifically, in the pure new product prediction scenario, the first characteristic information of the new product to be tested may be the price, specifications, origin, brand, category and other information of the new product to be tested. Multiple candidate products refer to products of the same type as the new product to be tested. Since products of the same type are more similar, the stability and accuracy of determining the reference products based on similarity can be further guaranteed.

[0090] In addition, the similarity between the first feature information of the new product to be tested and the first feature information of multiple candidate products is measured by a similarity algorithm, which can be Facebook AI Similarity Search (FAISS) and Siamese neteork (SBERT), or Word2Vec algorithm (Word to vector) and Word Mover's Distance (WMD), which is not limited in this application. The following introduces two methods for calculating similarity:

[0091] Method 1: SBERT is the abbreviation of Siamese network. Its sub-networks all use the BERT model (Bidirectional Encoder Representations from Transformers, BERT), and the two BERT models share parameters; FAISS is a C++-based library for indexing vectorized data and performing text similarity search. For each candidate product, the first feature information of the new product to be tested and the candidate product is input into the SBERT pre-trained model for feature extraction to obtain the feature vectors of the new product to be tested and the candidate product; based on the two sets of feature vectors, the FAISS algorithm is used to calculate the similarity between the two sets of features (that is, the similarity between the new product to be tested and the candidate product), and the candidate product with the highest similarity is selected as the reference product.

[0092] Method 2: Word2Vec is a natural language processing (NLP) technology for generating word vectors. It can convert words into vectors in a high-dimensional space so that semantically similar words are close in distance in the vector space. It has two structures: one is the skip-gram structure, which uses the intermediate words of the text information to predict the contextual neighboring word vector; the other is the continuous bag of words (CBOW) structure, which uses the contextual neighboring words of the text information to predict the intermediate word vector. Word2Vec uses the hidden layer weight values ​​produced during the pre-training network model to one-hot encode the first feature information to obtain the corresponding feature vector. Based on the feature vectors of the new product to be tested and the candidate product, the word shift distance algorithm is used to calculate the similarity between the two sets of vectors. The word shift distance algorithm is a benchmark algorithm applied to text similarity, which measures the shortest distance required for a word in one text to move to another text word in the semantic control space.

[0093] In the skip-gram structure, there is a skip-gram model that needs to maximize the conditional probability of predicting the contextual neighboring words given any intermediate word. The calculation formula is shown in the following formula (1):

[0094]

[0095] Among them, T represents the text length of the first feature information, t represents the time step, m identifies the time window, P represents the conditional probability, ω (t) The word with time step t, ω (t+j) Represents the word at time step t+j.

[0096] The conditional probability is defined by the softmax function and can be used to embed the model parameters into the loss function. The specific calculation formula is shown in the following formula (2):

[0097]

[0098] Among them, ω o Indicates the contextual neighboring words; ω c represents the middle word; ν represents the dictionary index set; o represents the index of the context neighboring word in the dictionary; c represents the index of the middle word in the dictionary; u represents the vector of the context neighboring word; v c Represents the vector of the middle word. P(ω o |ω c ) represents the conditional probability between the word at time step t and the word at time step (t+j).

[0099] The logarithm of the conditional probability in formula (2) is averaged, and the logarithm of the conditional probability in formula (2) is obtained by differentiation with respect to the intermediate word vector v c The specific calculation formula is shown in the following formula (3):

[0100]

[0101] Among them, u o The vector representing the contextual neighboring words; v c The vector representing the middle word; ω o Indicates the contextual neighboring words; ω c represents the middle word; ν represents the dictionary index set; o represents the index of the context neighboring word in the dictionary; c represents the index of the middle word in the dictionary, represents the gradient parameter.

[0102] The equivalent of minimizing the loss function is solved by the maximum likelihood method, and the gradient parameter calculated by the above formula (3) is implanted into the loss function. The specific calculation formula of minimizing the loss function is shown in the following formula (4):

[0103]

[0104] Among them, T represents the text length of the first feature information, t represents the time step, m identifies the time window, P represents the conditional probability, ω (t) The word with time step t, ω (t+j) Represents the word at time step t+j.

[0105] The specific calculation steps in the continuous bag of words (CBOW) structure are the same as those in the skip-gram structure.

[0106] For each candidate product, the first feature information of the new product to be tested and the candidate product is input into the Word2Vec pre-trained network model for feature extraction to obtain the word vectors of the new product to be tested and the candidate product. The cosine similarity between the word vectors of the new product to be tested and the reference product is calculated. The candidate product with the highest similarity is selected as the reference product.

[0107] In this application, the calculation formula of the price elasticity coefficient is shown in the following formula (5):

[0108]

[0109] Wherein, e represents the price elasticity coefficient, P represents the price of the reference commodity, ΔP represents the difference between the price of the reference commodity and the pricing of the new product to be tested; Q represents the historical sales information of the reference commodity, ΔQ represents the difference between the historical sales information of the reference commodity and the target sales information of the new product to be tested; dP represents the rate of change of the reference commodity price, and dQ represents the rate of change of the historical sales information of the reference commodity with the change of price.

[0110] The price elasticity coefficient of the reference product is calculated based on the historical sales information and price of the reference product; based on the elastic price coefficient, the difference ΔQ between the historical sales information of the reference product and the target sales information of the new product to be tested is calculated. The target sales information of the new product to be tested is obtained based on the historical sales information of the reference product and the difference ΔQ.

[0111] In the embodiments of the present application, since the price of fresh goods is the most important factor affecting their sales, changes in the prices of such goods will lead to huge differences in sales. Introducing a price elasticity coefficient in a pure new product prediction scenario to predict the target prediction information of the new products to be tested can reduce potential losses caused by overpricing of the new products to be tested.

[0112] In some embodiments, when the target prediction scenario is a seasonal new product prediction scenario, the present application uses the following method to predict the target sales information of the new product to be tested:

[0113] Obtain second feature information of multiple candidate products within a preset time range; select a reference product from the multiple candidate products based on the similarity between the second feature information of the new product to be tested and the second feature information of the multiple candidate products; determine a first adjustment coefficient based on the discount information of the reference product and the discount information of the new product to be tested; predict the target sales information of the new product to be tested based on the historical sales information of the reference product and the first adjustment coefficient.

[0114] Specifically, the multiple candidate products are products of the same type as the new product to be tested; the preset time range refers to the existing historical sales information of the new product to be tested that is more than 21 days from the current time. Of course, it can also be other time ranges, and this application does not make specific restrictions on this.

[0115] In the seasonal new product prediction scenario, the second characteristic information of the candidate products includes: the first characteristic information, the average sales ratio of the same type of products, holiday characteristics, activity discount characteristics, store sales orders, out-of-stock rate of the same type of products, etc. When selecting reference products from multiple candidate products, in addition to meeting the similarity requirements, if the new product to be tested is a strong seasonal new product, then on the basis of meeting the similarity, select the reference product of the same period in history; if the new product to be tested is a weak seasonal new product, then select the reference product whose last sales time is closest to the present.

[0116] The second feature information of the new product to be tested and multiple candidate products is normalized by max-min, and then the cosine similarity is calculated. The specific calculation formula is shown in the following formulas (6), (7), and (8):

[0117] A = (normalized feature 1 of the new product to be tested, normalized feature 2 of the new product to be tested, ..., normalized feature n of the new product to be tested) (6)

[0118] B = (normalized feature 1 of candidate product, normalized feature 2 of candidate product, ..., normalized feature n of candidate product) (7)

[0119]

[0120] Among them, A represents the normalized feature set of the new product to be tested; B represents the normalized feature set of the candidate products; similarity represents the similarity between the new product to be tested and the candidate products; and n represents the number of features in the first feature information.

[0121] The first adjustment coefficient can be the ratio of the discount information of the reference product to the discount information of the new product to be tested, or it can be the ratio between the pricing of the new product to be tested and the price of the reference product. For example, assuming that the historical sales information of the reference product is 11 units when the discount is 20% off, if the current discount of the new product to be tested is 95% off, then the target sales information of the new product to be tested is 11*0.8 / 0.95≈9.27 units.

[0122] In some embodiments, when the target prediction scenario is a new product prediction scenario with sales, the present application uses the following method to predict the target sales information of the new product to be tested:

[0123] If the shelf time of the new product to be tested is less than the first preset time, the second adjustment coefficient is determined based on the historical sales information of reference products of the same type as the new product to be tested; based on the sales growth coefficient of the new product to be tested within the first preset time, the historical sales information of the new product to be tested and the second adjustment coefficient, the target sales information of the new product to be tested is predicted.

[0124] If the shelf time of the new product to be tested is greater than the second preset time, a machine learning regression model is used to predict the target sales information of the new product to be tested based on the second feature information of the new product to be tested and the historical sales information of the new product to be tested.

[0125] Specifically, the first preset duration and the second preset duration can be set according to actual conditions, and the second preset duration is greater than the first preset duration.

[0126] For example, the new product period of the new product to be tested is generally 14 or 21 natural days, the first preset time period may be 50% of the new product period, the second preset time period may be 80% of the new product period, and the second preset time period is greater than the first preset time period.

[0127] The second adjustment coefficient includes: the holiday sales coefficient of the reference product and the activity coefficient of the new product to be tested, wherein the holiday sales coefficient of the reference product is the ratio of the holiday sales information of the reference product to the non-holiday sales information, and the activity coefficient of the new product to be tested is the ratio of the activity discount price pre-set on the forecast date of the new product to be tested to the non-activity price.

[0128] The sales growth coefficient is the ratio of the sales information of the new product to be tested on the Nth day after it was put on the shelves to the sales information on the first day after it was put on the shelves. If there is no holiday sales information in the existing historical sales information of the new product to be tested, the holiday sales information of the new product to be tested is restored based on the holiday sales coefficient of the reference product and the existing historical sales information of the new product to be tested. If there is no activity sales information in the existing historical sales information of the new product to be tested, the activity sales information of the new product to be tested is restored based on the activity coefficient of the new product to be tested and the existing historical sales information of the new product to be tested; based on the sales growth coefficient of the new product to be tested within the first preset time period, the restored holiday sales information or activity sales information, the target sales information of the new product to be tested is predicted.

[0129] In this application, see Figure 4 The third characteristic information includes: sales characteristics, product characteristics, forward warehouse characteristics of related stores, holiday characteristics, activity characteristics, etc., among which sales characteristics can be average sales; product characteristics include: price, average sales of similar products, etc.; forward warehouse characteristics of related stores include: forward warehouse order quantity, number of forward warehouse VIP members, etc.; holiday characteristics include: whether it is a holiday, holiday type, etc.; activity characteristics include: activity discount rate, activity type, etc.

[0130] See also Figure 5 The training process of the machine learning regression model in this application is as follows: construct samples and features based on the information of the same type of goods, use the constructed samples and features to train the machine learning regression model, and evaluate the training results.

[0131] The prediction process of the machine learning regression model in the present application is as follows: if the shelf time of the new product to be tested is greater than the second preset time, the third feature information and historical sales information of the new product to be tested are input into the trained machine learning regression model for prediction to obtain the predicted target sales information of the new product to be tested.

[0132] In the embodiments of the present application, full consideration is given to the fact that when the shelf time of the new product to be tested is less than the first preset time, the daily target sales information may fluctuate greatly, and after the shelf time is greater than the second preset time, the sales of the new product to be tested gradually tends to a stable state. According to the difference in shelf time, two sales prediction models for the new products to be tested are constructed, which improves the prediction accuracy of the target sales information of the new products to be tested in the prediction scenario of new products with sales, and further improves the interpretability of the target sales information prediction model of the new products to be tested.

[0133] In some embodiments, when the target prediction scenario is a reference new product prediction scenario, the present application uses the following method to predict the target sales information of the new product to be tested:

[0134] Based on the attribute information of the target store associated with the new product to be tested and the similarity between the attribute information of multiple candidate stores, a reference store is selected from multiple candidate stores; based on the difference information between the reference store and the target store, a third adjustment coefficient is determined; based on the historical sales information of the reference store associated with the new product to be tested and the third adjustment coefficient, the target sales information of the new product to be tested is predicted

[0135] Specifically, the attribute information of a store includes: the store's preset average daily order volume, the store's geographical location information, the store's average sales ratio information for each category of merchandise, etc. Among them, the store's geographical location information includes the longitude and latitude of the city where the store is located, the hierarchical relationship between the city and municipal divisions, etc.

[0136] The third adjustment coefficient can be any one of the following: a store scaling coefficient obtained based on the category average sales ratio of the target store and the reference store, a store conversion coefficient obtained based on the ratio of the average daily order volume of the reference store to the estimated daily order volume of the target store, and a store copy coefficient of a manually designated target store relative to the reference store.

[0137] The new product prediction scenario with reference can be further subdivided into: new product prediction sub-scenario based on sales volume of reference stores, new product prediction sub-scenario for newly opened stores, and new product full copy sub-scenario across stores.

[0138] In the sub-scenario of new product sales prediction in reference stores, the new product to be tested is not available in the target store. Based on the attribute information of the target store, a similarity algorithm is used to search for the reference store with the highest similarity to the target store and which has the new product to be tested available. The store scaling factor is determined based on the ratio between the average sales ratio of the same category of goods in the reference store and the average sales ratio of the same category of goods in the target store. The target sales information of the new product to be tested in the target store is predicted based on the historical sales information of the reference store associated with the new product to be tested and the store scaling factor.

[0139] In the new product prediction sub-scenario for newly opened stores, the target store is a newly opened store, and the appropriate reference store is determined by manual designation or similarity algorithm. The store conversion coefficient is determined based on the ratio between the average daily sales volume of the reference store and the preset daily sales volume of the target store. Based on the historical sales information associated with each new product to be tested in the reference store and the store conversion coefficient, the target sales information of each new product to be tested is predicted. The minimum predicted value of the target sales information of the new product to be tested is 1, and if it is less than 1, it is predicted as 1; since some new products to be tested in the target store are used for opening activities, the target sales information is directly set to a fixed increment and does not participate in the prediction.

[0140] In the embodiment of the present application, after obtaining the target predicted sales information of each new product to be tested, a predicted value is set for the new product to be tested whose target sales information is less than 1. For some new products to be tested that participate in specific activities such as opening activities, a fixed increment is set separately as the target sales information. This post-processing method further alleviates the problem of weak interpretability of the new product prediction model.

[0141] When the target store has been opened for more than three days, each new product to be tested in the target store will no longer predict the target sales information by referring to the historical sales information of the store associated with each new product to be tested.

[0142] In the sub-scenario of full copying of new products across stores, the target stores need to fully stock the new products to be tested. Based on the attribute information of the target stores, a similarity algorithm is used to search for reference stores with the highest similarity to the target stores. The date restriction range of the reference stores is set, and a store copy coefficient is manually specified. The copy coefficient can be dynamically adjusted according to the size of the store. Based on the historical sales information and store copy coefficients of the reference stores associated with the new products to be tested within the copy date restriction range, the target sales information of each new product to be tested is predicted.

[0143] If the reference store does not have the new product to be tested on the shelves, a separate prediction will be made for the new product to be tested.

[0144] If there are products overlapping with the new product to be tested in the reference store, the historical sales information of the overlapping products in the reference store is directly copied as the target sales information of the new product to be tested in the target store.

[0145] For example, see Figure 6 There are overlapping commodities (including one or more new products to be tested) between target store A and reference store B. Therefore, according to the store copy coefficient, the historical sales information of the overlapping commodities in reference store B is copied as the target sales information of the overlapping commodities in the target store. Reference store B also includes commodities that are not on the shelves of target store A. Therefore, according to the store copy coefficient, the historical sales information of the commodities that are not on the shelves of target store A in reference store B can also be copied as the target sales information of these commodities that are not on the shelves in the target store.

[0146] In an embodiment of the present application, when the target prediction scenario for the new product to be tested is a reference new product prediction scenario, multiple situations are further divided according to the differences in stores, and the target sales information of the new product to be tested is post-processed after obtaining the target sales information of the new product to be tested, which not only improves the interpretability of the sales prediction scheme for the new product to be tested, but also improves the accuracy of the new product sales prediction results.

[0147] In some embodiments, when the target prediction scenario is a substitute new product prediction scenario, the present application predicts the target sales information of the new product to be tested in the following manner:

[0148] The historical sales information of reference commodities that have a substitution relationship with the new product to be tested is obtained, and the target sales information of the new product to be tested is predicted based on the historical sales information of the reference commodities and a preset adjustment coefficient.

[0149] Specifically, the reference commodities that have a substitution relationship with the alternative new products and the preset adjustment coefficients are all manually specified.

[0150] In the embodiment of the present application, not only the characteristic information of the new product to be tested itself is effectively utilized, but also the attribute information of the target store associated with the new product to be tested and the characteristic information of the reference product are fully utilized, thereby improving the accuracy of the target sales information predicted for the new product to be tested in this scenario, thereby confirming the objectivity and interpretability of the prediction model. The explanatory method is shown in Table 1 below.

[0151] Table 1

[0152] New product prediction scenario Explanatory methods Pure new product prediction scenario Similarity + Reference Product + Price Elasticity Method Seasonal new product forecasting scenario Reference product historical sales range + reference product historical sales New product sales forecast scenario Historical sales + growth coefficient; historical sales + regression model Reference store sales new product forecast sub-scenario Reference store + reference store product sales + store scaling factor New product prediction sub-scenario for new stores Reference store + store conversion factor New product cross-store full copy sub-scenario Reference store + store copy coefficient + store copy date

[0153] Based on the same technical concept, the present application embodiment provides a structural diagram of a sales information prediction device, such as Figure 7 As shown, the sales information prediction device 700 includes:

[0154] The receiving module 701 is used to obtain the scene attribute information associated with the new product prediction scene corresponding to the new product to be tested.

[0155] The judgment module 702 is used to determine the target prediction scene corresponding to the new product to be tested from multiple new product prediction scenes based on the scene attribute information corresponding to the new product to be tested, and different new product prediction scenes correspond to different new product prediction models.

[0156] The processing module 703 is used to use the new product prediction model corresponding to the target prediction scenario to predict the target sales information of the new product to be tested.

[0157] Optionally, the scene attribute information includes: listing information;

[0158] The determination module 702 is specifically used for:

[0159] If the listing information indicates that the new product to be tested has been listed, a new product prediction scenario with sales volume among the multiple new product prediction scenarios is used as a target prediction scenario corresponding to the new product to be tested.

[0160] Optionally, the scene attribute information also includes: historical sales information;

[0161] The determination module 702 is further used for:

[0162] If the listing information indicates that the new product to be tested has not been listed, and the historical sales information indicates that the new product to be tested has historical sales, then the seasonal new product prediction scenario among the multiple new product prediction scenarios is used as the target prediction scenario corresponding to the new product to be tested.

[0163] Optionally, the scene attribute information further includes: reference sales information;

[0164] The determination module 702 is further used for:

[0165] If the listing information indicates that the new product to be tested has not been listed, the historical sales information indicates that the new product to be tested has no historical sales, and the reference sales information indicates that the new product to be tested has reference sales, then the reference new product prediction scenario among the multiple new product prediction scenarios is used as the target prediction scenario corresponding to the new product to be tested;

[0166] If the listing information indicates that the new product to be tested has not been listed, the historical sales information indicates that the new product to be tested has no historical sales, and the reference sales information indicates that the new product to be tested has no reference sales, then the pure new product prediction scenario among the multiple new product prediction scenarios is used as the target prediction scenario corresponding to the new product to be tested.

[0167] Optionally, the target prediction scenario is a pure new product prediction scenario; the processing module 703 is specifically used to:

[0168] The target prediction scenario is a pure new product prediction scenario, and based on the similarity between the first feature information of the new product to be tested and the first feature information of multiple candidate products, a reference product is selected from the multiple candidate products;

[0169] Based on the historical sales information and price elasticity coefficient of the reference product, the target sales information of the new product to be tested is predicted.

[0170] Optionally, the target prediction scenario is a seasonal new product prediction scenario; the processing module 703 is specifically used to:

[0171] The target prediction scenario is a seasonal new product prediction scenario, and second feature information of multiple candidate products within a preset time range is obtained;

[0172] Selecting a reference commodity from the plurality of candidate commodities based on the similarity between the second feature information of the new commodity to be tested and the second feature information of the plurality of candidate commodities;

[0173] Determining a first adjustment coefficient based on the discount information of the reference product and the discount information of the new product to be tested;

[0174] Based on the historical sales information of the reference product and the first adjustment coefficient, the target sales information of the new product to be tested is predicted.

[0175] Optionally, the target prediction scenario is an alternative new product prediction scenario;

[0176] The processing module 703 is specifically used for:

[0177] The target prediction scenario is a substitute new product prediction scenario, and the historical sales information of reference products that have a substitute relationship with the new product to be tested is obtained;

[0178] Based on the historical sales information of the reference product and the preset adjustment coefficient, the target sales information of the new product to be tested is predicted.

[0179] Optionally, the target prediction scenario is a new product prediction scenario with sales volume;

[0180] The processing module 703 is specifically used for:

[0181] The target prediction scenario is a prediction scenario for new products with sales volume. If the shelf time of the new product to be tested is less than the first preset time, a second adjustment coefficient is determined based on historical sales information of reference products of the same type as the new product to be tested.

[0182] Predicting target sales information of the new product to be tested based on the sales growth coefficient of the new product to be tested within the first preset time period, the historical sales information of the new product to be tested, and the second adjustment coefficient;

[0183] If the shelf time of the new product to be tested is longer than the second preset time, a machine learning regression model is used to predict the target sales information of the new product to be tested based on the third feature information of the new product to be tested and the historical sales information of the new product to be tested, and the second preset time is longer than the first preset time.

[0184] Optionally, the target prediction scenario is the reference new product prediction scenario;

[0185] The processing module 703 is specifically used for:

[0186] The target prediction scenario is the reference new product prediction scenario, and based on the similarity between the attribute information of the target store associated with the new product to be tested and the attribute information of multiple candidate stores, a reference store is selected from the multiple candidate stores;

[0187] Determining a third adjustment coefficient based on difference information between the reference store and the target store;

[0188] Based on the historical sales information of the reference store associated with the new product to be tested and the third adjustment coefficient, the target sales information of the new product to be tested is predicted.

[0189] Based on the same technical concept, the embodiment of the present application provides a computer device, such as Figure 8 As shown, it includes at least one processor 801 and a memory 802 connected to the at least one processor. The specific link medium between the processor 801 and the memory 802 is not limited in the embodiment of the present application. Figure 8 For example, the processor 801 and the memory 802 are connected via a bus. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0190] In the embodiment of the present application, the memory 802 stores instructions executed by at least one processor 801, and the at least one processor 801 can execute the steps of the above-mentioned sales information prediction method by executing the instructions stored in the memory 802.

[0191] Among them, the processor 801 is the control center of the computer device, and can use various interfaces and lines to connect various parts of the computer device, and realize sales information prediction by running or executing instructions stored in the memory 802 and calling data stored in the memory 802. Optionally, the processor 801 may include one or more processing modules, and the processor 801 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 801. In some embodiments, the processor 801 and the memory 802 may be implemented on the same chip, and in some embodiments, they may also be implemented separately on independent chips.

[0192] Processor 801 can be a general-purpose processor, such as a central processing unit (CPU), a digital signal processor, an application-specific integrated circuit (ASIC), a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, and can implement or execute the methods, steps and logic block diagrams disclosed in the embodiments of the present application. A general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware processor for execution, or can be executed by a combination of hardware and software modules in the processor.

[0193] The memory 802 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory 802 may include at least one type of storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory, a random access memory (Random Access Memory, RAM), a static random access memory (Static Random Access Memory, SRAM), a programmable read-only memory (Programmable Read Only Memory, PROM), a read-only memory (Read Only Memory, ROM), an electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, EEPROM), a magnetic memory, a disk, an optical disk, etc. The memory 802 is any other medium that can be used to carry or store a desired program code in the form of an instruction or data structure and can be accessed by a computer device, but is not limited thereto. The memory 802 in the embodiment of the present application can also be a circuit or any other device that can realize a storage function, for storing program instructions and / or data.

[0194] Based on the same inventive concept, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program that can be executed by a computer device. When the program runs on the computer device, the computer device executes the steps of the above-mentioned sales information prediction method.

[0195] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0196] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0197] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0198] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0199] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A sales information prediction method, characterized in that: include: Obtaining scene attribute information associated with the new product prediction scene corresponding to the new product to be tested; Based on the scene attribute information corresponding to the new product to be tested, determining a target prediction scene corresponding to the new product to be tested from multiple new product prediction scenes, where different new product prediction scenes correspond to different new product prediction models; The new product prediction model corresponding to the target prediction scenario is used to predict the target sales information of the new product to be tested.

2. The method according to claim 1, characterized in that The scene attribute information includes: listing information; The determining, based on the scene attribute information corresponding to the new product to be tested, a target prediction scene corresponding to the new product to be tested from a plurality of new product prediction scenes includes: If the listing information indicates that the new product to be tested has been listed, a new product prediction scenario with sales volume among the multiple new product prediction scenarios is used as a target prediction scenario corresponding to the new product to be tested.

3. The method according to claim 2, characterized in that Also includes: The scene attribute information also includes: historical sales information; If the listing information indicates that the new product to be tested has not been listed, and the historical sales information indicates that the new product to be tested has historical sales, then the seasonal new product prediction scenario among the multiple new product prediction scenarios is used as the target prediction scenario corresponding to the new product to be tested.

4. The method according to claim 3, characterized in that Also includes: The scene attribute information also includes: reference sales information; If the listing information indicates that the new product to be tested has not been listed, the historical sales information indicates that the new product to be tested has no historical sales, and the reference sales information indicates that the new product to be tested has reference sales, then the reference new product prediction scenario among the multiple new product prediction scenarios is used as the target prediction scenario corresponding to the new product to be tested; If the listing information indicates that the new product to be tested has not been listed, the historical sales information indicates that the new product to be tested has no historical sales, and the reference sales information indicates that the new product to be tested has no reference sales, then the pure new product prediction scenario among the multiple new product prediction scenarios is used as the target prediction scenario corresponding to the new product to be tested.

5. The method according to claim 1, characterized in that The target prediction scenario is a pure new product prediction scenario; the new product prediction model corresponding to the target prediction scenario is used to predict the target sales information of the new product to be tested, including: Selecting a reference commodity from the plurality of candidate commodities based on the similarity between the first feature information of the new commodity to be tested and the first feature information of a plurality of candidate commodities; Based on the historical sales information and price elasticity coefficient of the reference product, the target sales information of the new product to be tested is predicted.

6. The method according to claim 1, characterized in that The target prediction scenario is a seasonal new product prediction scenario; the new product prediction model corresponding to the target prediction scenario is used to predict the target sales information of the new product to be tested, including: Acquire second feature information of multiple candidate commodities within a preset time range; Selecting a reference commodity from the plurality of candidate commodities based on the similarity between the second feature information of the new commodity to be tested and the second feature information of the plurality of candidate commodities; Determining a first adjustment coefficient based on the discount information of the reference product and the discount information of the new product to be tested; Based on the historical sales information of the reference product and the first adjustment coefficient, the target sales information of the new product to be tested is predicted.

7. The method according to claim 1, characterized in that The target prediction scenario is a substitute new product prediction scenario; the new product prediction model corresponding to the target prediction scenario is used to predict the target sales information of the new product to be tested, including: Obtaining historical sales information of reference products that have a substitution relationship with the new product to be tested; Based on the historical sales information of the reference product and the preset adjustment coefficient, the target sales information of the new product to be tested is predicted.

8. The method according to claim 1, characterized in that The target prediction scenario is a new product prediction scenario with sales volume; the new product prediction model corresponding to the target prediction scenario is used to predict the sales volume information of the new product to be tested, including: If the shelf time of the new product to be tested is less than the first preset time, determining a second adjustment coefficient based on historical sales information of reference products of the same type as the new product to be tested; Predicting target sales information of the new product to be tested based on the sales growth coefficient of the new product to be tested within the first preset time period, the historical sales information of the new product to be tested, and the second adjustment coefficient; If the shelf time of the new product to be tested is longer than the second preset time, a machine learning regression model is used to predict the target sales information of the new product to be tested based on the third feature information of the new product to be tested and the historical sales information of the new product to be tested, and the second preset time is longer than the first preset time.

9. The method according to claim 1, characterized in that The target prediction scenario is the reference new product prediction scenario; and the new product prediction model corresponding to the target prediction scenario is used to predict the target sales information of the new product to be tested, including: Based on the similarity between the attribute information of the target store associated with the new product to be tested and the attribute information of multiple candidate stores, a reference store is selected from the multiple candidate stores; Determining a third adjustment coefficient based on difference information between the reference store and the target store; Based on the historical sales information of the reference store associated with the new product to be tested and the third adjustment coefficient, the target sales information of the new product to be tested is predicted.

10. A transaction processing device, characterized in that: The device comprises: A receiving module, used to obtain scene attribute information associated with a new product prediction scene corresponding to the new product to be tested; A judgment module, used to determine a target prediction scene corresponding to the new product to be tested from multiple new product prediction scenes based on scene attribute information corresponding to the new product to be tested, wherein different new product prediction scenes correspond to different new product prediction models; The processing module is used to use the new product prediction model corresponding to the target prediction scenario to predict the target sales information of the new product to be tested.

11. A computer device, characterized in that: include: Memory for storing computer programs; A processor is used to call the computer program stored in the memory, and execute the method according to any one of claims 1 to 9 according to the obtained program.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program, and when the program is run on a computer, the computer is enabled to implement the method according to any one of claims 1 to 9.