CN112288484A

CN112288484A - Commodity automatic pricing method and device, electronic equipment and storage medium

Info

Publication number: CN112288484A
Application number: CN202011187013.6A
Authority: CN
Inventors: 冯楚; 王攀
Original assignee: Hangzhou Netease Zaigu Technology Co Ltd
Current assignee: Hangzhou Netease Zaigu Technology Co Ltd
Priority date: 2020-10-30
Filing date: 2020-10-30
Publication date: 2021-01-29

Abstract

The embodiment of the invention provides an automatic commodity pricing method. The method comprises the following steps: obtaining pricing element information of a target commodity, wherein the pricing element information is divided according to corresponding types and at least comprises one of commodity attribute information, commodity cost information and marketing action information; and inputting the pricing element information into a price prediction model to determine the price information of the target commodity, wherein the price prediction model is obtained by training commodity historical data of associated commodities, and the associated commodities and the target commodity belong to the same brand. The method can effectively improve the automatic commodity pricing precision and effectively reduce the computing resource consumption of the automatic commodity pricing. In addition, the embodiment of the invention provides an automatic commodity pricing device, an electronic device and a computer readable storage medium.

Description

Commodity automatic pricing method and device, electronic equipment and storage medium

Technical Field

The embodiment of the invention relates to the technical field of information processing, in particular to an automatic commodity pricing method, an automatic commodity pricing device, electronic equipment and a storage medium.

Background

This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. The description herein is not admitted to be prior art by inclusion in this section.

With the rapid development of computer technology, the application of commodity price prediction through an information processing mode becomes more and more extensive. The commodity price prediction is beneficial to improving the service operation efficiency in the electronic commerce field and improving the intelligent degree of commodity pricing in the electronic commerce field.

In implementing the present disclosure, the inventors found that in the related art, when performing automatic pricing of a product, price prediction for the target product is performed by combining historical price data of the target product and market environment information.

However, for some target commodities, the actual market environment may differ significantly from the ideal market environment in which the commodity is completely free to compete, which may result in a high error rate for price predictions for the target commodity; and the commodity price prediction is carried out by combining with market environment information, and the defects of low prediction efficiency, large computing resource consumption and high pricing cost exist.

Disclosure of Invention

In the prior art, when automatic pricing of commodities is performed, price prediction for the target commodities is performed by combining historical price data of the target commodities and market environment information. However, for some target commodities, the actual market environment may differ significantly from the ideal market environment in which the commodity is completely free to compete.

Therefore, in the prior art, the problem that the error rate of the determined price information of the target commodity is high probably exists by combining the historical price data of the target commodity and the market environment information; and the commodity price prediction is carried out by combining with the market environment information, so that the problems of low prediction efficiency, high calculation resource consumption and high pricing cost exist, and the process is very annoying.

Therefore, an improved method for automatically pricing commodities is highly needed to realize a method for determining commodity price information, which has high pricing accuracy, high reference value, improved pricing efficiency and low consumption of computing resources.

In this context, embodiments of the present invention are intended to provide a method, an apparatus, an electronic device, and a storage medium for automatically pricing goods.

In a first aspect of an embodiment of the present invention, a method for automatically pricing a commodity is provided, including obtaining pricing factor information of a target commodity, where the pricing factor information is divided according to corresponding types and includes at least one of commodity attribute information, commodity cost information, and marketing action information; and inputting the pricing element information into a price prediction model to determine the price information of the target commodity, wherein the price prediction model is obtained by training commodity historical data of associated commodities, and the associated commodities and the target commodity belong to the same brand.

In one embodiment of the present invention, the price prediction model is obtained by training product history data of the associated product, and includes: and taking the historical pricing information in the commodity history data as a label value, and performing primary feature construction based on the commodity history data to obtain a corresponding original feature set.

In another embodiment of the present invention, the method further comprises: calculating the correlation coefficient of each feature in the original feature set and the label value; and expanding or screening the original feature set based on the correlation coefficient to obtain a first target feature set for secondary feature construction.

In another embodiment of the present invention, the secondary feature configuration includes a trial run screening step: inputting data corresponding to the first target feature set into the price prediction model for model running, and determining the precision gain coefficient of each feature in the first target feature set to the price prediction model according to the running result; screening the first target feature set based on the correlation coefficient and the precision gain coefficient to obtain a second target feature set; wherein the features in the second target feature set correspond to the pricing factor information for the associated item.

In another embodiment of the present invention, the secondary feature construction further includes a fusion expansion step: and fusing features corresponding to different types of pricing factor information according to the correlation coefficient and the precision gain coefficient, and adding new features formed by fusion to the second target feature set to expand the second target feature set.

In another embodiment of the present invention, the step of model running further comprises: and determining the MSE score of the price prediction model according to the running result, wherein the MSE score obtained by the first running is used as the MSE reference score.

In another embodiment of the present invention, the method further comprises repeating the run-out screening step and the fusion expansion step, and re-determining the MSE score of the price prediction model until the corresponding MSE score is no longer improved and is higher than the reference MSE score compared to the previous time; iteratively obtaining a final version of the second set of target features based on the MSE reference score.

In another embodiment of the present invention, the price prediction model is obtained by training product history data of the associated product, and includes: and randomly dividing the commodity history data into a plurality of commodity history data subsets corresponding to different time periods based on the time dimension, and training the price prediction model by using the commodity history data subsets of different time periods respectively.

In another embodiment of the present invention, the method further comprises an effect evaluation step of: training the commodity historical data subset with the time period before to generate the price prediction model, and determining the price prediction value of the associated commodity at a certain time period after based on the corresponding price prediction model; determining the actual price of the associated commodity at a certain time period according to the commodity historical data subset corresponding to the certain time period; and evaluating the prediction effect of the price prediction model based on the price prediction value and the actual price of the associated commodity in a certain period of time.

In another embodiment of the present invention, the method further comprises: correcting the price information of the target commodity determined by the price prediction model based on a preset pricing strategy; the pricing policies include at least one of a lowest gross rate pricing policy, a bundled pricing policy, and an odd numbered pricing policy.

In another embodiment of the present invention, the method further comprises: the price prediction model is periodically updated based on the Hive T +1 model.

In another embodiment of the present invention, the correlation coefficient

Wherein X represents the characteristic value of each characteristic in the original characteristic set,

the mean of the feature values representing all the features in the original feature set, Y represents the actual pricing of the respective associated goods,

a mean value representing the actual pricing of all associated goods;

precision gain factor associated with a feature

Therein, MSE₁When data corresponding to the characteristic is not input into the price prediction model for model test run, the first MSE score, MSE of the price prediction model₂And when the data corresponding to the characteristics are input into the price prediction model for model test, the second MSE score of the price prediction model is obtained.

In another embodiment of the present invention, the article attribute information includes at least one of article color information, article size information, and article category information; the commodity cost information at least comprises one of purchasing cost information, tax cost information and operation cost information; the marketing action information at least comprises one of commodity shelving information, commodity price adjusting information and commodity sales promotion information.

In a second aspect of the embodiments of the present invention, there is provided an automatic commodity pricing device, including an obtaining module, configured to obtain pricing factor information of a target commodity, where the pricing factor information is divided according to corresponding types and includes at least one of commodity attribute information, commodity cost information, and marketing action information; and the processing module is used for inputting the pricing factor information into a price prediction model to determine the price information of the target commodity, wherein the price prediction model is obtained by training commodity historical data of associated commodities, and the associated commodities and the target commodity belong to the same brand.

In one embodiment of the invention, the processing module comprises: and the first processing submodule is used for taking the historical pricing information in the commodity historical data as a label value and carrying out primary feature construction based on the commodity historical data so as to obtain a corresponding original feature set.

In another embodiment of the present invention, the processing module further comprises: the second processing submodule is used for calculating the correlation coefficient of each feature in the original feature set and the label value; and the third processing submodule is used for expanding or screening the original feature set based on the correlation coefficient so as to obtain a first target feature set for secondary feature construction.

In yet another embodiment of the present invention, the processing module includes: the trial run screening submodule is used for inputting data corresponding to the first target characteristic set into the price prediction model to perform model trial run, and determining the precision gain coefficient of each characteristic in the first target characteristic set to the price prediction model according to a trial run result; screening the first target feature set based on the correlation coefficient and the precision gain coefficient to obtain a second target feature set; wherein the features in the second target feature set correspond to the pricing factor information for the associated item.

In another embodiment of the present invention, the processing module further comprises: and the fusion expansion submodule is used for fusing the characteristics corresponding to the pricing factor information of different types according to the correlation coefficient and the precision gain coefficient, and adding the new characteristics formed by fusion to the second target characteristic set so as to expand the second target characteristic set.

In another embodiment of the present invention, the run screening sub-module further comprises: and the MSE score determining unit is used for determining the MSE score of the price prediction model according to the trial run result, wherein the MSE score obtained by the first trial run is used as the MSE reference score.

In another embodiment of the present invention, the determining the MSE score of the price prediction model by the trial screening submodule and the fusion expansion submodule is repeated until the corresponding MSE score is no longer improved than the MSE score of the previous time and is higher than the reference MSE score; iteratively obtaining a final version of the second set of target features based on the MSE reference score.

In another embodiment of the present invention, the processing module is configured to randomly divide the product history data into a plurality of product history data subsets corresponding to different time periods based on a time dimension, and train the price prediction model with the plurality of product history data subsets corresponding to different time periods.

In another embodiment of the present invention, the processing module further includes an effect evaluation sub-module, configured to train and generate the price prediction model with the commodity history data subset with a time period earlier, and determine a price prediction value of the associated commodity at a later time period based on the corresponding price prediction model; determining the actual price of the associated commodity at a certain time period according to the commodity historical data subset corresponding to the certain time period; and evaluating the prediction effect of the price prediction model based on the price prediction value and the actual price of the associated commodity in a certain period of time.

In another embodiment of the present invention, the processing module further includes a price information optimization sub-module, configured to modify the price information of the target product determined by the price prediction model based on a preset pricing strategy; the pricing policies include at least one of a lowest gross rate pricing policy, a bundled pricing policy, and an odd numbered pricing policy.

In another embodiment of the present invention, the processing module further comprises a model update sub-module: the price prediction model is periodically updated based on the Hive T +1 model.

In another embodiment of the present invention, the correlation coefficient

a mean value representing the actual pricing of all associated goods;

precision gain factor associated with a feature

In a third aspect of embodiments of the present invention, there is provided an electronic apparatus, comprising: one or more processors; memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of any of the first aspects.

In a fourth aspect of embodiments of the present invention, there is provided a computer-readable storage medium storing computer-executable instructions for implementing the method of any one of the first aspect when executed.

According to the automatic commodity pricing method and device provided by the embodiment of the invention, the problem that the determined price information has a large error based on the historical price data and the market environment information of the target commodity can be effectively solved, and meanwhile, because the pricing element information has the advantages of small data volume and simple data acquisition mode compared with the market environment information, the embodiment of the invention can also effectively reduce the calculation resource consumption of automatic commodity pricing, and is beneficial to improving the automatic commodity pricing efficiency.

Drawings

The above and other objects, features and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description read in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which:

fig. 1 schematically shows an exemplary system architecture of an automatic commodity pricing method and an apparatus thereof according to an embodiment of the present invention;

FIG. 2 schematically illustrates a flow chart of a method for automatic pricing of goods according to an embodiment of the invention;

FIG. 3 schematically illustrates a flow chart of another method for automatic pricing of items according to an embodiment of the invention;

FIG. 4 schematically shows a schematic diagram of a training process of a price prediction model according to an embodiment of the invention;

FIG. 5 schematically shows a schematic diagram of a model effect evaluation method according to an embodiment of the invention;

FIG. 6 schematically illustrates a schematic diagram of an automatic merchandise pricing process according to an embodiment of the invention;

FIG. 7 schematically illustrates a program product for enabling automatic pricing of goods according to an embodiment of the invention;

FIG. 8 schematically illustrates a block diagram of an automatic merchandise pricing device according to an embodiment of the invention; and

FIG. 9 schematically illustrates a computing device that may implement automatic pricing of items according to an embodiment of the invention.

In the drawings, the same or corresponding reference numerals indicate the same or corresponding parts.

Detailed Description

The principles and spirit of the present invention will be described with reference to a number of exemplary embodiments. It is understood that these embodiments are given solely for the purpose of enabling those skilled in the art to better understand and to practice the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

As will be appreciated by one skilled in the art, embodiments of the present invention may be embodied as a system, apparatus, device, method, or computer program product. Accordingly, the present disclosure may be embodied in the form of: entirely hardware, entirely software (including firmware, resident software, micro-code, etc.), or a combination of hardware and software.

According to the embodiment of the invention, a method, a device, electronic equipment and a computer readable medium for automatically pricing commodities are provided.

In this document, it is to be understood that any number of elements in the figures are provided by way of illustration and not limitation, and any nomenclature is used for differentiation only and not in any limiting sense.

The principles and spirit of the present invention are explained in detail below with reference to several representative embodiments of the invention.

Summary of The Invention

The inventor finds that in the related art, when automatic pricing of a product is performed, price prediction for the target product is performed by combining historical price data of the target product and market environment information. However, for some target commodities, the actual market environment may differ significantly from the ideal market environment in which the commodity is completely free to compete, which may result in a high error rate for price predictions for the target commodity; because the market environment information has the characteristics of multiple data types, large data amount and non-centralized data distribution, the market environment information is combined to predict the commodity price, and the market environment information has the defects of low prediction efficiency, high calculation resource consumption and high pricing cost.

Moreover, for some brands with own modes of goods design schemes, goods brands and goods sales channels, market replaceability is relatively weak, and goods of the brands have certain free brand pricing space. Compared with the ideal market environment of completely free competition of commodities, the commodities are mostly in the monopolized market or the short market, and the commodity brand side has certain price control. In the related art, the automatic pricing of the commodities based on the historical pricing information, the market environment information, the historical sales information and the like of the commodities cannot be realized, the commodity pricing based on brand premium cannot be realized, and the pricing requirement of the commodities with a certain brand free pricing space cannot be met.

In the existing practice of the industry, pricing scenes based on brand premium are usually given by a data system to assist decision information, or business personnel collect required assist decision information, and the business personnel need to analyze and judge by themselves based on experience and the acquired assist decision information to give final pricing of commodities. The pricing operation flow is a fracturing flow, and the manual operation also introduces decision deviation and pricing fluctuation.

Therefore, on the basis of reasonable historical pricing, the embodiment of the invention discovers the pricing strategy implicit in the commodity historical pricing, trains and obtains the price prediction model for the automatic commodity pricing by utilizing the commodity historical data containing the brand pricing strategy, and realizes the automation of the commodity pricing. In addition, the process does not depend on the pricing information of similar commodities in the market and the feedback of market environment information on commodity pricing, can effectively improve the automatic commodity pricing efficiency, and can be well suitable for automatic commodity pricing with certain brand autonomy and brand free pricing space.

Having described the general principles of the invention, various non-limiting embodiments of the invention are described in detail below.

Application scene overview

First, an exemplary system architecture of the method and the apparatus for automatically pricing commodities according to an embodiment of the present invention is described in detail with reference to fig. 1.

As shown in fig. 1, a system architecture 100 according to this embodiment may include

business servers

101, 102, 103, a network 104, and a processing device 105. The network 104 is used to provide a medium for communication links between the

traffic servers

101, 102, 103 and the processing device 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.

The

business servers

101, 102, and 103 store pricing factor information of the target product, and the pricing factor information is divided according to corresponding types and includes at least one of product attribute information, product cost information, and marketing action information. The

service servers

101, 102, and 103 may be servers in a service platform that sells target products, or may be servers for storing pricing factor data of target products. The present embodiment does not limit the types of the service servers.

The processing device 105 is a device capable of automatically pricing a product, and may be, for example, a processor capable of performing information processing and data calculation. The processing device 105 acquires pricing factor information of the target commodity from the

business servers

101, 102, 103, and then inputs the pricing factor information into a price prediction model to determine price information of the target commodity, wherein the price prediction model is obtained by training commodity history data of an associated commodity, and the associated commodity and the target commodity belong to the same brand.

It should be noted that the automatic pricing method for goods provided by the embodiment of the present invention may be generally executed by the processing device 105. Accordingly, the automatic merchandise pricing apparatus provided by the embodiment of the invention may be generally disposed in the processing device 105. The method for automatically pricing commodities provided by the embodiment of the present invention may also be executed by other processing devices or processing device clusters different from the processing device 105 and capable of communicating with the

service servers

101, 102, 103 and/or the processing device 105. Accordingly, the automatic commodity pricing apparatus provided by the embodiment of the present invention may also be disposed in other processing devices or processing device clusters different from the processing device 105 and capable of communicating with the

service servers

101, 102, 103 and/or the processing device 105.

It should be understood that the number of traffic servers, networks, and processing devices in fig. 1 is merely illustrative. There may be any number of traffic servers, networks, and processing devices, as desired for implementation.

Exemplary method

In the following, in conjunction with the application scenario of fig. 1, a method for automatically pricing commodities according to an exemplary embodiment of the present invention is described with reference to fig. 2 and 3. It should be noted that the above application scenarios are merely illustrated for the convenience of understanding the spirit and principles of the present invention, and the embodiments of the present invention are not limited in this respect. Rather, embodiments of the present invention may be applied to any scenario where applicable.

Fig. 2 schematically shows a flowchart of an automatic commodity pricing method according to an embodiment of the present invention.

As shown in fig. 2, the method is applied to a terminal including at least one application program that can be executed. The method 200 may include steps S210-S220.

Step S210, pricing factor information of the target commodity is obtained, wherein the pricing factor information is divided according to corresponding types and at least comprises one of commodity attribute information, commodity cost information and marketing action information.

According to the embodiment of the invention, when price prediction aiming at the target commodity is required, pricing element information of the target commodity is obtained, wherein the pricing element information is element information capable of influencing pricing of the target commodity. The target commodities can comprise entity commodities and virtual commodities, pricing element information is divided according to corresponding types, and specifically, the pricing element information can comprise commodity attribute information, commodity cost information, marketing action information and the like.

Specifically, the article attribute information may include, for example, article color information, article model information, article size information, article category information, article performance information, and the like. The goods cost information may include, for example, production cost information, purchasing cost information, tax cost information, transportation cost information, storage cost information, operation cost information, etc., and the operation cost information may include, for example, promotion cost information, electronic commerce operation cost information, etc. The marketing action information may include, for example, item listing information, item adjustment information, item promotion information, and the like, and the item promotion information may include, for example, full reduction information, full gift information, coupon information, package coupon information, and the like. The category and content of the obtained pricing element information can be determined according to different service scenes and actual needs, and the application is not limited herein.

And step S220, inputting the pricing element information into a price prediction model to determine the price information of the target commodity, wherein the price prediction model is obtained by training commodity historical data of the associated commodity, and the associated commodity and the target commodity belong to the same brand.

According to the embodiment of the invention, the pricing element information of the target commodity is input into the preset price prediction model, so that the price prediction of the target commodity is realized by using the price prediction model. The price prediction model is obtained by training commodity history data of the associated commodity, the associated commodity and the target commodity belong to the same brand, and the commodity history data comprises historical pricing information of the associated commodity and other information capable of influencing historical pricing of the associated commodity. In the case of performing model training using the product history data of the related product, the feature set corresponding to the product history data can be obtained by performing feature construction using other information that can affect the history pricing of the related product, with the history pricing information in the product history data being a label value. And performing model training in a supervised regression mode by using the characteristic set corresponding to the commodity historical data to obtain a price prediction model. The associated commodity and the target commodity belong to the same brand, and the associated commodity and the target commodity under the same brand are usually operated by the same or similar brand strategies, so that the brand values contained in the associated commodity and the target commodity are not greatly different. In one embodiment of the present invention, the associated goods are shoes of brand a, and the target goods are clothing belonging to brand a, so that by using the solution of one embodiment of the present invention, a model can be trained through the goods history data of the shoes of brand a, so as to predict the price of the clothing of brand a. It should be noted that the associated product and the target product may be the same or similar products, or different products, and as long as the associated product and the target product belong to the same brand, the scheme in the embodiment of the present invention may be adopted to predict the price of the target product.

Because the commodity historical data of the associated commodity contains the brand value information of the brand, the price prediction model in the embodiment is adopted to predict the price information of the target commodity, so that the predicted price information can reflect the brand value of the brand, the commodity to be online can keep a consistent brand operation strategy of the brand, and a merchant can be helped to earn excess profits brought by brand overprice, and the problem that the existing price prediction model is difficult to accurately predict the commodity price containing the brand value is solved. On the other hand, merchants with high brand value are often in monopolized or semi-monopolized market positions in market competition, which causes that the market environment of the merchants does not conform to the ideal state and basic assumption of a complete competitive market, so that the existing price prediction model based on the ideal state is difficult to provide reasonable pricing for commodities launched by the merchants, and serious prediction deviation even causes huge profit loss for the merchants.

The price prediction model disclosed in one embodiment of the invention is adopted to predict the price of the target commodity, and the price prediction model is obtained by training commodity historical data of the associated commodity, so that on one hand, the training data comes from the interior of the brand, and the data can be independently controlled and easily obtained; on the other hand, the commodity history data reflects the real market performance of the associated commodity of the brand in the actual facing market environment, so that the reasonable price of the target commodity can be determined before the commodity is online by adopting the price prediction model in the embodiment, so that the corresponding price is closer to the real market environment while realizing high brand premium.

Optionally, after determining the price information of the target commodity, the price information may be modified based on a preset pricing strategy, wherein the pricing strategy includes at least one of a lowest gross interest rate pricing strategy, a bundled pricing strategy and an odd pricing strategy. The lowest gross profit rate pricing strategy specifically may be to preset a lowest gross profit rate, and when the determined price information cannot meet the preset lowest gross profit rate, modify the price information according to a lowest gross profit rate standard. The binding pricing strategy specifically can be a pricing strategy made for the combined commodity, and when the target commodity is the combined commodity or needs to be sold in combination with other commodities, the preset binding pricing strategy is used for correcting the price information. The odd pricing strategy may specifically be a pricing strategy in which a high price sensitive to the user is set to be low and a low price insensitive to the user is set to be high, and illustratively, the predicted price of the target product determined by the price prediction model is 111 yuan, and the predicted price of the target product is modified from 111 yuan to 109 yuan by the odd pricing strategy.

And after the price information determined by the price prediction model is corrected, obtaining the recommended pricing of the target commodity, and returning the recommended pricing of the target commodity to the service system through the service interface so that the service system can finally determine the actual pricing for the target commodity according to the received recommended pricing. Or the service providing system determines a expenditure willingness index, a brand preference index, a performance demand index and the like of the user aiming at the target commodity according to the received recommended pricing, so as to adjust the operation strategy aiming at the target commodity.

The technical scheme of the embodiment of the invention provides an automatic commodity pricing method, which is characterized in that pricing element information of a target commodity is obtained, wherein the pricing element information is divided according to corresponding types and at least comprises one of commodity attribute information, commodity cost information and marketing action information; and inputting the pricing element information into a price prediction model to determine the price information of the target commodity, wherein the price prediction model is obtained by training commodity historical data of the associated commodity, and the associated commodity and the target commodity belong to the same brand. The price information of the target commodity is determined by utilizing a price prediction model, and the price prediction model is obtained by training commodity historical data of associated commodities belonging to the same brand, so that the price prediction of the target commodity can not be strongly influenced by market environment information; the embodiment of the invention can carry out automatic commodity pricing without depending on market environment information, but can carry out automatic pricing aiming at the target commodity by obtaining the pricing element information of the target commodity based on the pricing element information, and compared with the market environment information, the pricing element information has more simplified data content and simpler obtaining mode, thereby effectively reducing the workload of data obtaining and data processing, effectively improving the forecasting efficiency of commodity price forecasting, being beneficial to reducing the computing resource consumption of the commodity price forecasting and further being beneficial to realizing the automatic commodity pricing with low cost and high efficiency.

FIG. 3 schematically illustrates a flow chart of another method for automatic pricing of items according to an embodiment of the invention.

As shown in FIG. 3, the training process of the price prediction model utilized in step S220 may include steps S310-S330.

Step S310, the historical pricing information in the commodity historical data is used as a label value, and primary feature construction is carried out on the basis of the commodity historical data to obtain a corresponding original feature set.

According to the embodiment of the present invention, in the case of performing the price prediction model training, the product history data of a plurality of related products is acquired, for example, the product history data of a plurality of related products for a given sku, where sku is the minimum stock unit of a product and the same product of different specifications corresponds to different skus, may be acquired. The product history data includes historical pricing information and other information that can affect historical pricing of the associated product, and may include, for example, product attribute information, product cost information, historical marketing information, and the like. After the product history data is obtained, word segmentation and encoding processing may be performed on the character string data in the product history data. Illustratively, the commodity history data includes "green cup price is 12 yuan", and the character string data "green cup", "price" and "yuan" need to be subjected to word segmentation and encoding processing, so that the character string data is processed into character features.

Performing word segmentation and coding processing on character string data in the commodity history data to realize processing of other information capable of influencing historical pricing of related commodities in the commodity history data into character features, taking the part of the character features as training features, taking historical pricing information in the commodity history data as a label value, integrating the label value and the training features by taking sku as a unique key, and realizing primary feature construction based on the commodity history data to obtain a corresponding original feature set.

Alternatively, when the price prediction model is obtained by training using the product history data of the associated product, the product history data may be randomly divided into a plurality of product history data subsets corresponding to different time periods based on the time dimension, and then the price prediction model may be trained with the plurality of product history data subsets corresponding to different time periods, respectively.

Step S320, calculating a correlation coefficient between each feature in the original feature set and the tag value, and expanding or screening the original feature set based on the correlation coefficient to obtain a first target feature set.

According to the embodiment of the present invention, a statistical method is used to determine a correlation coefficient between each feature in the original feature set and a tag value, that is, a correlation coefficient between each feature in the original feature set and historical pricing information is determined, where the statistical method may include, for example, a descriptive statistical method, a sampling inference method, a time series analysis method, an exponential analysis method, and the like, and the correlation coefficient may specifically be expressed as:

representing the average of the actual pricing of all associated goods.

When the correlation coefficient of a certain feature in the original feature set and the tag value is higher than a preset threshold value, it can be judged that the correlation degree between the feature and the tag value is high, and the feature can be used as the key attention data of the subsequent feature construction. Illustratively, the correlation coefficients between different features and tag values are shown by the following table.

After determining the correlation coefficient between each feature in the original feature set and the tag value, the original feature set is expanded or screened based on the correlation coefficient to obtain a first target feature set. Specifically, the target feature with the correlation coefficient higher than the preset threshold is screened from the original feature set, or the commodity history data corresponding to other features related to the target feature is obtained in an expanding mode, the correlation coefficient between the feature in the commodity history data obtained in the expanding mode and the tag value is continuously judged, and if the correlation coefficient is higher than the preset threshold, the corresponding feature is screened as the target feature, so that the original feature set is expanded. And forming a first target feature set by the target features obtained after the expansion or screening treatment.

And expanding or screening the original feature set based on the correlation coefficient of each feature in the original feature set and the tag value to obtain a first target feature set. The correlation coefficient between each feature in the first target feature set and the label value is higher than the preset threshold value, so that the high correlation degree between each feature in the first target feature set and the historical price information of the associated commodity is effectively ensured, and the accuracy of the constructed first target feature set is effectively ensured.

Step S330, performing secondary feature construction on the first target feature set to obtain a second target feature set, wherein features in the second target feature set correspond to pricing factor information of the associated commodities.

According to the embodiment of the invention, the secondary feature construction of the first target feature set comprises a trial run screening step, specifically, data corresponding to the first target feature set is input into a price prediction model for model trial run, the precision gain coefficient of each feature in the first target feature set to the price prediction model is determined according to the trial run result, then the first target feature set is screened based on the correlation coefficient and the precision gain coefficient to obtain a second target feature set, wherein the features in the second target feature set correspond to pricing element information of associated commodities.

In addition, the model run step further comprises: and determining the MSE score of the price prediction model according to the running result, wherein the MSE score obtained by the first running is used as the MSE reference score. Specifically, data corresponding to the first target feature set are input into the price prediction model for model running, and the MSE score of the price prediction model is determined according to a running result. The model running result is the price information of the associated commodity output by the price prediction model, namely the predicted price of the price prediction model for the associated commodity. The MSE score is specifically mean square error and is used for measuring the deviation between the price predicted value and the true value, and the specific calculation formula is as follows:

wherein, P_iIs the predicted price, P 'of the ith associated good'_iIs the actual price of the ith associated item.

After data corresponding to any feature in the first target feature set is input into the price prediction model for model test, a first MSE score of the price prediction model associated with the feature can be obtained. And determining the precision gain coefficient of the characteristic to the price prediction model by combining the second MSE score of the price prediction model when the data corresponding to the characteristic is not input into the price prediction model. I.e. the precision gain factor associated with any feature

Therein, MSE₁When data corresponding to the characteristic is not input into the price prediction model for model test run, the first MSE score, MSE of the price prediction model₂When inputting the data corresponding to the characteristic into the price prediction model for model test, the second MSE of the price prediction model is obtainedAnd (4) dividing. Illustratively, for a p5 signature, MSE₁When data corresponding to the characteristics of p0, p1, p2, p3 and p4 are input into the price prediction model for model test run, the first MSE score, MSE, of the price prediction model₂When data corresponding to p0, p1, p2, p3, p4 and p5 characteristics are input into the price prediction model for model test, the second MSE score of the price prediction model and the precision gain coefficient associated with the p5 characteristics

After data corresponding to each feature in the first target feature set is input into the price prediction model for model trial run, the MSE score of the price prediction model associated with each feature is determined, and the MSE score can represent the deviation degree between the predicted price and the actual price of the associated commodity, namely the prediction accuracy of the price prediction model. The MSE score of the price prediction model is determined, the precision gain coefficient associated with each feature is determined based on the MSE score, the association degree between each feature and the historical pricing information of the associated commodity can be clearly and accurately obtained, and the pricing strategy implied in the historical pricing information of the associated commodity can be better explored. And performing secondary feature construction aiming at the first target feature set based on the correlation coefficient and the precision gain coefficient associated with each feature, so that the accuracy of a feature construction result is favorably ensured, and the prediction precision of a price prediction model is favorably improved.

Optionally, the secondary feature construction further includes a fusion expansion step, specifically, features corresponding to different types of pricing factor information are fused according to the correlation coefficient and the precision gain coefficient, and new features formed by fusion are added to the second target feature set to expand the second target feature set. And combining the correlation coefficient, the precision gain coefficient and the service experience, and fusing the features of which the correlation coefficient and/or the precision gain coefficient are higher than a preset threshold and correspond to different types of pricing factor information by using modes of combination calculation, grouping statistics and the like to obtain new training features. Optionally, the new training features obtained by fusion may be input into the price prediction model to perform model running, so as to obtain a model running result, and a precision gain coefficient for the new training features is determined according to the model running result, and when the precision gain coefficient is higher than a preset threshold, the new training features are added into the second target feature set.

Optionally, when it is determined that the accuracy gain coefficient for the new training feature is higher than the preset threshold, the new training feature and other features may be further fused and expanded to further expand the second target feature set, and the second target feature set which is as complete as possible is constructed, so that a price prediction model with higher prediction accuracy is obtained through training.

For example, when finding that the correlation coefficient and the precision gain coefficient related to the cost-based features are high in model run, it may be determined that the historical pricing of the associated goods includes a cost-based premium strategy, and at this time, the cost-based features may be fused and expanded in a manner of grouping fusion, linear combination, polynomial combination, proportional combination, ranking coding features, outlier fusion, and the like, and new features formed by fusion are added to the second target feature set. Or when the accuracy gain coefficient of the brand features is found to be high in the model run-in, the historical pricing of the associated commodities can be judged to include a premium strategy based on brand information, and at the moment, the brand features such as commodity creators, brand positioning crowds and the like can be secondarily constructed to obtain new features after fusion and expansion.

And repeating the model running step and the fusion expansion step, and repeatedly determining the MSE score of the price prediction model. And when the MSE score of the price prediction model is not improved any more than the MSE reference score in the previous time and is higher than the MSE reference score, stopping the model running step and the fusion and expansion step, and thus obtaining the final version of the second target feature set in an iterative mode based on the MSE reference score. The features in the second target feature set correspond to pricing factor information of the associated good, i.e. the features in the second target feature set correspond to pricing policy information of the associated good. When the MSE score of the price prediction model is not improved any more than the MSE reference score in the previous time and is higher than the MSE reference score, the price prediction model can be well close to the actual price value aiming at the price prediction value of the associated commodity, the price prediction model can realize stable and high-precision commodity price prediction, and the price prediction model obtained through the last iteration can be used for actual automatic commodity pricing service.

FIG. 4 schematically shows a schematic diagram of a training process of a price prediction model according to an embodiment of the present invention, and as shown in FIG. 4, the training process may include 401-405 processes.

In step 401, the historical pricing information in the product history data is used as a label value, and the initial feature structure is performed on other data except the historical pricing information to obtain an original feature set composed of a plurality of training features.

In 402, a correlation coefficient between each feature in the original feature set and the tag value is determined, and then the original feature set is expanded or screened based on the correlation coefficient to obtain a first target feature set.

In the 403 process, secondary feature construction is performed on the features in the first target feature set, specifically, data corresponding to the first target feature set is input into the price prediction model to perform model run, and iteration is performed to obtain an MSE score of the price prediction model and obtain an accuracy gain coefficient associated with each feature in the first target feature set.

In the process 404, feature construction is performed on the features in the first target feature set based on the correlation coefficient and the accuracy gain coefficient, the feature construction includes screening and/or fusion expansion processing, and a second target feature set in the process 405 is obtained, wherein the features in the second target feature set correspond to pricing factor information of the associated commodity.

After a price prediction model meeting the precision requirement is obtained through the supervised regression mode training, pricing service can be provided through a mode of providing a service interface. Specifically, the system architecture of the embodiment of the method may include an algorithm service module and a pricing service module, the algorithm service module is responsible for updating the algorithm model and calculating the automatic pricing, and the pricing service module is responsible for providing the automatic commodity pricing service to the outside. The pricing service module obtains a pricing service request from the business system by providing an application program interface for the business system, then carries out price prediction operation aiming at the target commodity through the algorithm service module, and returns price information of the target commodity to the business system through the pricing service module.

Specifically, a service person inputs a designated sku through a service system, and the pricing service module acquires pricing element information of a target commodity of the designated sku in real time, specifically, the pricing element information of the target commodity can be acquired from different service subsystems in real time. And then, inputting the pricing element information into the algorithm service module, calculating by using a price prediction model in the algorithm service module to obtain pricing information of the target commodity, and then returning the pricing information to the service system through the pricing service module. In order to improve the automatic commodity pricing efficiency, pricing element information of target commodities with different skus can be maintained to databases such as mysql and the like based on real-time algorithms such as Flink and the like for query during commodity pricing service.

Feature construction and algorithm model training belong to calculation-intensive operations, and the computing capacity is generally flexibly improved by means of algorithms such as hadoop and k8 s. The pricing service module realizes the automatic commodity pricing service by providing an application program interface for the business system. The algorithm service module and the pricing service module are separately deployed, so that the safety and the stability of the automatic commodity pricing system are guaranteed.

Optionally, after the price prediction model is obtained through training, a prediction effect evaluation step can be performed. Specifically, when the model training is performed using the product history data of the associated product, the product history data is randomly divided into a plurality of product history data subsets corresponding to different time periods based on the time dimension, and then the price prediction model is trained using the plurality of product history data subsets corresponding to different time periods. Therefore, the price prediction model can be trained and generated by the commodity historical data subset with the front time period, and the price prediction value of the related commodity at a certain time period can be determined based on the corresponding price prediction model; determining the actual price of the related goods in a certain time period according to the goods history data subset corresponding to the certain time period; and evaluating the prediction effect of the price prediction model based on the price prediction value and the actual price of the associated commodity at a certain later time period.

Exemplarily, fig. 5 schematically shows a schematic diagram of a model effect evaluation method according to an embodiment of the present invention, as shown in fig. 5, a price prediction model is trained using commodity history sub data of commodities of 1 and 2 months, a price prediction value of a commodity of 4 months is determined based on the corresponding price prediction model, an actual price of the commodity of 4 months is determined according to the commodity history sub data of the commodity of 4 months, and a prediction accuracy of the price prediction model is determined according to the price prediction value and the actual price of the commodity of 4 months. Or training a price prediction model by using the commodity history subdata of the commodities in the 1 month, the 2 months and the 3 months, determining a price prediction value of the commodity in the 5 months by using the price prediction model, and then determining the prediction precision of the price prediction model based on the price prediction value and the actual price of the commodity in the 5 months.

Optionally, after the price prediction model is obtained through training, a model updating step may be further performed, specifically, the price prediction model may be periodically updated based on a Hive T +1 model, where T +1 refers to an update period that an update time of the algorithm model is delayed by one day relative to a service periodicity. When the model is updated, an algorithm characteristic set can be constructed based on Hive T +1, and updating training for the price prediction model can be performed by using the constructed algorithm characteristic set.

Specifically, the target feature set is constructed based on the basic data in the data warehouse at regular time and in full quantity every day, and specifically, the primary feature construction and the secondary feature construction can be performed on the basic data to obtain a second target feature set corresponding to the basic data. The basic data may be, for example, actual transaction data of the associated goods in the business system. And then, segmenting the second target feature set obtained in the previous step to obtain a plurality of second target feature subsets, and inputting data corresponding to each second target feature subset into the price prediction model for training so as to realize iterative updating aiming at the price prediction model.

Fig. 6 is a schematic diagram schematically illustrating an automatic commodity pricing process according to an embodiment of the present invention, and as shown in fig. 6, after a pricing service request from a business system is obtained, pricing element information of a target commodity is obtained in real time based on the pricing service request, and then automatic commodity pricing calculation is performed through a price prediction model to obtain price information of the target commodity, and the price information is returned to the business system. The price prediction model can be periodically updated based on Hivet +1, specifically, training features are constructed based on T +1, and an algorithm model is updated based on T +1, so that the price prediction model is updated.

According to the embodiment of the invention, the price prediction model is obtained by training with the commodity history data of the associated commodity belonging to the same brand as the target commodity, so that when the price information of the target commodity is determined by using the price prediction model, the automatic commodity pricing calculation can be carried out based on the historical pricing information and the brand pricing strategy, the price prediction for the target commodity can be not strongly dependent on market environment information, the price prediction model is obtained by training with the commodity history data of the associated commodity belonging to the same brand, the accuracy of automatic commodity pricing can be effectively improved, the price system of the commodity of the same brand can be kept stable, and strategic profit control can be realized; the obtained pricing element information has the advantages of small data volume and simple obtaining mode, so the embodiment is also beneficial to reducing the computing resource consumption of automatic commodity pricing and saving the pricing cost of automatic commodity pricing.

Exemplary Medium

Having described the method of an exemplary embodiment of the present invention, a computer-readable storage medium of an exemplary embodiment of the present invention is described next with reference to fig. 7. The computer-readable storage medium stores computer-executable instructions, which when executed by the processing unit, are used for implementing the automatic commodity pricing method applied to the terminal in any one of the above method embodiments.

In some possible embodiments, aspects of the present invention may also be implemented in the form of a program product comprising program code for causing a computing device to perform the steps in the method for automatic pricing of items according to various exemplary embodiments of the present invention described in the above section "exemplary method" of this specification, when the program product is run on the computing device, for example, the computing device may perform step S210 as shown in fig. 2: obtaining pricing element information of the target commodity, wherein the pricing element information is divided according to corresponding types and at least comprises one of commodity attribute information, commodity cost information and marketing action information; step S220: and inputting the pricing element information into a price prediction model to determine the price information of the target commodity, wherein the price prediction model is obtained by training commodity historical data of the associated commodity, and the associated commodity and the target commodity belong to the same brand.

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.

As shown in fig. 7, a program product 70 for automatic pricing of goods according to an embodiment of the invention is depicted, which may employ a portable compact disc read only memory (CD-ROM) and include program code, and may be run on a computing 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.

A 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 any of a variety of 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).

Exemplary devices

Having described the medium of an exemplary embodiment of the present invention, an automatic commodity pricing apparatus according to an exemplary embodiment of the present invention will be described with reference to fig. 8 and 9.

Fig. 8 schematically shows a block diagram of an automatic commodity pricing apparatus according to an embodiment of the present invention.

As shown in fig. 8, the automatic merchandise pricing apparatus 800 may include an obtaining module 810 and a processing module 820. The processing means may perform the method as described in the method embodiments section.

Specifically, the obtaining module 810 is configured to obtain pricing factor information of the target product, where the pricing factor information is divided according to corresponding types and includes at least one of product attribute information, product cost information, and marketing action information; and the processing module 820 is used for inputting the pricing factor information into a price prediction model to determine the price information of the target commodity, wherein the price prediction model is obtained by training the commodity historical data of the associated commodity, and the associated commodity and the target commodity belong to the same brand.

In yet another embodiment of the present invention, a processing module comprises: the trial run screening submodule is used for inputting data corresponding to the first target characteristic set into the price prediction model for model trial run and determining the precision gain coefficient of each characteristic in the first target characteristic set to the price prediction model according to a trial run result; screening the first target feature set based on the correlation coefficient and the precision gain coefficient to obtain a second target feature set; wherein the features in the second target feature set correspond to pricing factor information for the associated good.

In another embodiment of the present invention, the processing module further comprises: and the fusion expansion submodule is used for fusing the features corresponding to the pricing factor information of different types according to the correlation coefficient and the precision gain coefficient, and adding the new features formed by fusion to the second target feature set so as to expand the second target feature set.

In another embodiment of the invention, the MSE score of the price prediction model is determined repeatedly by the trial run screening submodule and the fusion expansion submodule until the corresponding MSE score is not improved any more than the MSE score of the previous time and is higher than the reference MSE score; based on the MSE reference score, to iteratively obtain a final version of the second set of target features.

In another embodiment of the present invention, the processing module is configured to randomly segment the product history data into a plurality of product history data subsets corresponding to different time periods based on the time dimension, and train the price prediction model with the plurality of product history data subsets corresponding to different time periods.

In another embodiment of the present invention, the processing module further includes an effect evaluation sub-module, configured to train and generate a price prediction model with the commodity history data subset with a previous time period, and determine a price prediction value of the associated commodity at a later time period based on the corresponding price prediction model; determining the actual price of the related goods in a certain time period according to the goods history data subset corresponding to the certain time period; and evaluating the prediction effect of the price prediction model based on the price prediction value and the actual price of the associated commodity at a certain later time period.

In another embodiment of the invention, the correlation coefficient

a mean value representing the actual pricing of all associated goods;

precision gain factor associated with a feature

In another embodiment of the present invention, the article attribute information includes at least one of article color information, article size information, and article category information; the commodity cost information at least comprises one of purchasing cost information, tax cost information and operation cost information; the marketing action information at least comprises one of commodity shelving information, commodity price adjusting information and commodity promotion information.

Since each functional module of the automatic commodity pricing device 800 of the exemplary embodiment of the present invention corresponds to the steps of the above-described exemplary embodiment of the automatic commodity pricing method, it is not described herein again.

Exemplary computing device

Having described the method, medium, and apparatus of exemplary embodiments of the present invention, a computing device of exemplary embodiments of the present invention for implementing the method of automatic pricing of items of the present invention will now be described with reference to FIG. 9.

The embodiment of the invention also provides the computing equipment. As will be appreciated by one skilled in the art, aspects of the present invention may be embodied as an apparatus, 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.

In some possible embodiments, a computing device according to the present invention may include at least one processing unit, and at least one memory unit. Wherein the storage unit stores program code that, when executed by the processing unit, causes the processing unit to perform the steps of the method for automatic pricing of merchandise according to various exemplary embodiments of the present invention described in the above section "exemplary method" of the present specification. For example, the processing unit may perform operation S210 as shown in fig. 2: obtaining pricing element information of the target commodity, wherein the pricing element information is divided according to corresponding types and at least comprises one of commodity attribute information, commodity cost information and marketing action information; operation S220: and inputting the pricing element information into a price prediction model to determine the price information of the target commodity, wherein the price prediction model is obtained by training commodity historical data of the associated commodity, and the associated commodity and the target commodity belong to the same brand.

A computing device 900 that may perform automatic pricing of items according to the present invention is described below with reference to fig. 9. The computing device 900 shown in FIG. 9 is only one example and should not be taken to limit the scope of use and functionality of embodiments of the present invention.

As shown in fig. 9, computing device 900 is embodied in a general purpose computing device. Components of computing device 900 may include, but are not limited to: the at least one processing unit 910, the at least one memory unit 920, and a bus 930 that couples various system components including the memory unit 920 and the processing unit 910.

Bus 930 represents one or more of any of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures.

The storage unit 920 may include readable media in the form of volatile memory, such as a Random Access Memory (RAM)921 and/or a cache memory 922, and may further include a Read Only Memory (ROM) 923.

Storage unit 920 may also include programs/utilities 925 having a set (at least one) of program modules 924, such program modules 924 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.

Computing device 900 may also communicate with one or more external devices 940 (e.g., keyboard, pointing device, bluetooth device, etc.), with one or more devices that enable a user to interact with computing device 900, and/or with any devices (e.g., router, modem, etc.) that enable computing device 900 to communicate with one or more other computing devices. Such communication may occur via input/output (I/O) interface 950. Moreover, computing device 900 may also 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 network adapter 960. As shown, the network adapter 960 communicates with the other modules of the computing device 900 via the bus 930. It should be appreciated that although not shown in the figures, other hardware and/or software modules may be used in conjunction with computing device 900, 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.

It should be noted that although in the above detailed description several units/modules or sub-units/modules of the apparatus are mentioned, such a division is merely exemplary and not mandatory. Indeed, the features and functionality of two or more of the units/modules described above may be embodied in one unit/module according to embodiments of the invention. Conversely, the features and functions of one unit/module described above may be further divided into embodiments by a plurality of units/modules.

Moreover, while the operations of the method of the invention are depicted in the drawings in a particular order, this does not require or imply that the operations must be performed in this particular order, or that all of the illustrated operations 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 down into multiple step executions.

While the spirit and principles of the invention have been described with reference to several particular embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, nor is the division of aspects, which is for convenience only as the features in such aspects may not be combined to benefit. The invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. An automatic commodity pricing method, comprising:

obtaining pricing element information of a target commodity, wherein the pricing element information is divided according to corresponding types and at least comprises one of commodity attribute information, commodity cost information and marketing action information;

and inputting the pricing element information into a price prediction model to determine the price information of the target commodity, wherein the price prediction model is obtained by training commodity historical data of associated commodities, and the associated commodities and the target commodity belong to the same brand.

2. The method of claim 1, wherein the price prediction model is trained from commodity biographical data of associated commodities, comprising:

and taking the historical pricing information in the commodity history data as a label value, and performing primary feature construction based on the commodity history data to obtain a corresponding original feature set.

3. The method of claim 2, further comprising:

calculating the correlation coefficient of each feature in the original feature set and the label value;

and expanding or screening the original feature set based on the correlation coefficient to obtain a first target feature set for secondary feature construction.

4. The method of claim 3, wherein the secondary feature construction comprises a run screening step:

inputting data corresponding to the first target feature set into the price prediction model for model running, and determining the precision gain coefficient of each feature in the first target feature set to the price prediction model according to the running result;

screening the first target feature set based on the correlation coefficient and the precision gain coefficient to obtain a second target feature set; wherein the features in the second target feature set correspond to the pricing factor information for the associated item.

5. The method of claim 4, wherein the secondary feature construction further comprises, a fusion augmentation step:

and fusing features corresponding to different types of pricing factor information according to the correlation coefficient and the precision gain coefficient, and adding new features formed by fusion to the second target feature set to expand the second target feature set.

6. The method of claim 5, wherein the step of model commissioning further comprises:

and determining the MSE score of the price prediction model according to the running result, wherein the MSE score obtained by the first running is used as the MSE reference score.

7. The method of claim 6, wherein the run-out screening step and the fusion expansion step are repeated and the MSE scores of the price prediction model are re-determined until the respective MSE scores are no longer elevated compared to the previous time and are above a baseline MSE score;

iteratively obtaining a final version of the second set of target features based on the MSE reference score.

8. An automatic commodity pricing device, comprising:

the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring pricing element information of a target commodity, and the pricing element information is divided according to corresponding types and at least comprises one of commodity attribute information, commodity cost information and marketing action information;

and the processing module is used for inputting the pricing factor information into a price prediction model to determine the price information of the target commodity, wherein the price prediction model is obtained by training commodity historical data of associated commodities, and the associated commodities and the target commodity belong to the same brand.

9. An electronic device, comprising:

one or more processors;

a memory for storing one or more programs,

wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of any of claims 1-7.

10. A computer-readable storage medium storing computer-executable instructions for implementing the method of any one of claims 1 to 7 when executed.