CN115860815A

CN115860815A - Merchant preference distribution method, device, equipment and medium

Info

Publication number: CN115860815A
Application number: CN202211487798.8A
Authority: CN
Inventors: 程锦东; 陈思念; 李洲; 薛振宇
Original assignee: Industrial and Commercial Bank of China Ltd ICBC
Current assignee: Industrial and Commercial Bank of China Ltd ICBC
Priority date: 2022-11-22
Filing date: 2022-11-22
Publication date: 2023-03-28

Abstract

The disclosure provides a merchant preferential allocation method, and relates to the field of artificial intelligence. The method comprises the following steps: carrying out merchant feature clustering on the N first merchants and the M second merchants to obtain S clustering clusters; obtaining the distribution historical characteristics of each second merchant in a first cluster according to the historical characteristics of at least one first merchant in the first cluster, wherein the first cluster is any one of the S clusters; inputting the distribution historical characteristics and the merchant characteristics of each second merchant into a pre-trained yield prediction model to obtain the predicted yield of each second merchant; and carrying out merchant preferential distribution according to the predicted income rate set and/or the historical income rate set of the M second merchants. The disclosure also provides a merchant offer apparatus, device, storage medium and program product.

Description

Merchant preference distribution method, device, equipment and medium

Technical Field

The present disclosure relates to the field of artificial intelligence, and more particularly, to a method, an apparatus, a device, a medium, and a program product for merchant offer distribution.

Background

The merchant preferential distribution can refer to the preferential treatment of the financial institution or the e-commerce platform for the fee such as the rate, the handling fee or the service fee of the merchant. Machine learning techniques can model and learn from a large number of existing samples and predict unknown samples. Machine learning techniques can be applied for merchant offer distribution.

The mainstream method at present mainly focuses on modeling and analyzing the current merchant characteristics of the merchant in the process of merchant preferential allocation. The merchant may have a significant increase in the amount of passenger or deal than before if only the current value of the feature is inaccurate at this time. It is also possible that some merchants, although some current merchant features are considerable, are limited by the type of sale, are not sensitive to offers, and it is not reasonable to lean too many offer resources to those merchants. Therefore, it is a problem to be solved at present that a new merchant benefit allocation method is provided to achieve reasonable allocation of merchant benefits.

Disclosure of Invention

In view of the foregoing, the present disclosure provides a merchant offer distribution method, apparatus, device, medium, and program product.

In one aspect of the embodiments of the present disclosure, a merchant offer distribution method is provided, including: carrying out merchant characteristic clustering on N first merchants and M second merchants to obtain S clustering clusters, wherein merchant characteristics are obtained according to inherent attribute data of the merchants, the N first merchants comprise merchants participating in historical preferential activities, the M second merchants comprise merchants not participating in the historical preferential activities, and N, M and S are integers greater than or equal to 1 respectively; obtaining distribution historical characteristics of each second merchant in a first cluster according to historical characteristics of at least one first merchant in the first cluster, wherein the first cluster is any one of the S clusters, and the historical characteristics are obtained according to data of the at least one first merchant in the process of participating in the historical preferential event; inputting the distribution historical characteristics and the merchant characteristics of each second merchant into a pre-trained yield prediction model to obtain the predicted yield of each second merchant, wherein the yield prediction model is obtained by training according to the merchant characteristic set, the historical characteristic set and the historical yield set of the N first merchants; and carrying out merchant preferential distribution according to the predicted income rate set and/or the historical income rate set of the M second merchants.

According to an embodiment of the present disclosure, before training the rate of return prediction model, the method includes: obtaining a candidate historical rate of return set of each first merchant of the N first merchants, wherein the candidate historical rate of return set comprises at least one sub rate of return, and the at least one sub rate of return corresponds to at least one historical discount joined by each first merchant one by one; distributing a time attenuation factor to each sub-rate of return in the at least one sub-rate of return according to the time sequence among the at least one historical preferential activity; and obtaining the historical profitability of each first merchant according to each sub-profitability and the corresponding time attenuation factor.

According to an embodiment of the present disclosure, before training the rate of return prediction model, the method includes: acquiring a candidate feature set of each first merchant in the N first merchants, wherein the candidate feature set includes a candidate merchant feature and a candidate history feature of each first merchant, the candidate merchant feature of each first merchant includes at least one merchant sub-feature, and the candidate history feature of each first merchant includes at least one history sub-feature; calculating the information gain of each merchant sub-feature and each history sub-feature in the candidate feature set; and screening each merchant sub-feature and each history sub-feature according to the information gain by using a decision tree algorithm to obtain the merchant feature set and the history feature set.

According to an embodiment of the present disclosure, includes: acquiring a historical income record of each first merchant, wherein the historical income record comprises income obtained by each first merchant participating in the historical preferential event; classifying the N first commercial tenants according to the historical income records; wherein the calculating the information gain of each merchant sub-feature and each historical sub-feature in the candidate feature set comprises: and calculating the information gain of each merchant sub-feature and each history sub-feature in the candidate feature set according to the classification results of the N first merchants.

According to the embodiment of the disclosure, before inputting the distribution history characteristics and merchant characteristics of each second merchant into the pre-trained yield prediction model, training the yield prediction model further includes: inputting N training samples into the yield prediction model to be trained, wherein the N training samples are obtained according to a merchant feature set, a historical feature set and a historical yield set of the N first merchants, and each training sample comprises a merchant feature, a historical feature and a historical yield of each first merchant; processing the merchant characteristics and the historical characteristics of each first merchant through the yield prediction model to obtain the predicted yield of each first merchant; and updating the parameters of the yield prediction model according to the difference degree between the predicted yield and the historical yield of each first merchant.

According to an embodiment of the present disclosure, the performing, according to the predicted revenue rate sets and/or the historical revenue rate sets of the M second merchants, merchant offer distribution includes: and performing merchant preferential distribution on the N first merchants and the M second merchants according to the predicted yield rate of each second merchant and the historical yield rate of each first merchant.

According to an embodiment of the present disclosure, the performing merchant offer distribution for the N first merchants and the M second merchants includes: under the condition that the total income amount of the N first merchants and the M second merchants is maximum, determining preferential allocation amount of each merchant of the N first merchants and the M second merchants; and calculating the total income amount according to the income rate and the preferential allocation amount of each merchant, wherein the income rate of each merchant comprises a predicted income rate or a historical income rate.

According to an embodiment of the present disclosure, further comprising: under at least one constraint condition, obtaining the maximum value of the total income amount by using an optimization algorithm; wherein the at least one constraint includes at least one of: the total preferential distribution amount of the N first commercial tenants and the M second commercial tenants is smaller than or equal to a first preset value; the total preferential distribution amount of each cluster in the S cluster clusters is smaller than or equal to a corresponding second preset value; the number of total merchants in the N first merchants and the M second merchants is less than or equal to a third preset value and greater than or equal to a fourth preset value; the number of the single merchants in each cluster participating in the preferential categories is smaller than or equal to a fifth preset value and larger than or equal to a sixth preset value.

Another aspect of the embodiments of the present disclosure provides a merchant offer distribution apparatus, including: the merchant distance module is used for carrying out merchant characteristic clustering on N first merchants and M second merchants to obtain S clustering clusters, wherein merchant characteristics are obtained according to inherent attribute data of the merchants, the N first merchants comprise merchants already participating in historical preferential activities, the M second merchants comprise merchants not participating in the historical preferential activities, and N, M and S are integers which are larger than or equal to 1 respectively; the characteristic distribution module is used for obtaining distribution historical characteristics of each second commercial tenant in a first cluster according to historical characteristics of at least one first commercial tenant in the first cluster, wherein the first cluster is any one of the S cluster, and the historical characteristics are obtained according to data of the at least one first commercial tenant in the process of participating in the historical preferential activities; the prediction yield rate module is used for inputting the distribution historical characteristics and the merchant characteristics of each second merchant into a pre-trained yield rate prediction model to obtain the prediction yield rate of each second merchant, wherein the yield rate prediction model is obtained by training according to the merchant characteristic set, the historical characteristic set and the historical yield rate set of the N first merchants; and the discount distribution module is used for carrying out merchant discount distribution according to the predicted income rate set and/or the historical income rate set of the M second merchants.

Another aspect of the disclosed embodiments provides an electronic device, including: one or more processors; a storage device to store 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 perform the method as described above.

Yet another aspect of the embodiments of the present disclosure provides a computer-readable storage medium having stored thereon executable instructions, which when executed by a processor, cause the processor to perform the method as described above.

Yet another aspect of the disclosed embodiments provides a computer program product comprising a computer program that when executed by a processor implements the method as described above.

One or more of the above embodiments have the following advantageous effects: in contrast to revenue prediction based directly on current merchant characteristics, embodiments of the present disclosure consider historical attribute characteristics of merchants. The first commercial tenant and the second commercial tenant are clustered into S cluster clusters by using a clustering method, and on the basis of the first commercial tenant which participates in the historical preferential event, distribution historical characteristics are obtained for the second commercial tenant which belongs to the same cluster, so that a pre-trained yield prediction model is expanded and applied to the commercial tenant which does not participate in the preferential event. And predicting yield before distributing the benefits, and screening merchants and distributing the benefits by using the model prediction result and/or the historical yield set, so that the benefit return of the development of the benefits is improved.

Drawings

The foregoing and other objects, features and advantages of the disclosure will be apparent from the following description of embodiments of the disclosure, which proceeds with reference to the accompanying drawings, in which:

fig. 1 schematically illustrates an application scenario diagram of a merchant offer distribution method according to an embodiment of the present disclosure;

FIG. 2 schematically illustrates a flow chart of a merchant offer distribution method according to an embodiment of the present disclosure;

FIG. 3 schematically illustrates a flow diagram of feature screening according to an embodiment of the present disclosure;

FIG. 4 schematically illustrates a flow chart for calculating an information gain according to an embodiment of the present disclosure;

FIG. 5 schematically illustrates a flow chart for obtaining historical profitability in accordance with an embodiment of the present disclosure;

FIG. 6 schematically shows a flow diagram for training a profitability prediction model according to an embodiment of the present disclosure;

FIG. 7 schematically illustrates a flow diagram of a merchant offer distribution method according to another embodiment of the present disclosure;

fig. 8 is a block diagram schematically illustrating a structure of a merchant offer distribution apparatus according to an embodiment of the present disclosure;

FIG. 9 schematically illustrates a block diagram of an electronic device suitable for implementing a merchant offer distribution method in accordance with an embodiment of the present disclosure.

Detailed Description

Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood that the description is illustrative only and is not intended to limit the scope of the present disclosure. In the following detailed description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the disclosure. It may be evident, however, that one or more embodiments may be practiced without these specific details. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present disclosure.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The terms "comprises," "comprising," and the like, as used herein, specify the presence of stated features, steps, operations, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, or components.

All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs, unless otherwise defined. It is noted that the terms used herein should be interpreted as having a meaning that is consistent with the context of this specification and should not be interpreted in an idealized or overly formal sense.

Where a convention analogous to "A, B and at least one of C, etc." is used, in general such a construction is intended in the sense one having skill in the art would understand the convention (e.g., "a system having at least one of A, B and C" would include, but not be limited to, systems that have a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.).

By taking a card swiping consumption example, a signing merchant is taken as a channel for card swiping consumption of a cardholder, and is an object of key attention of a financial institution order receiving service. Reasonable preferential distribution is provided for the merchants at irregular intervals, the activity of the merchants can be better activated, the viscosity of the financial institutions to the merchants is improved, and the profit is increased. However, the currently mainstream methods still have deficiencies in the preferential distribution to merchants. The existing mainstream distribution scheme focuses on the current characteristics of the merchants, such as passenger flow volume, daily transaction amount of the merchants and the like for modeling analysis, and the condition of unreasonable preferential distribution exists.

In contrast to revenue prediction based directly on current merchant characteristics, embodiments of the present disclosure consider historical attribute characteristics of merchants. The first commercial tenant and the second commercial tenant are clustered into S cluster clusters by using a clustering method, and on the basis of the first commercial tenant which participates in the historical preferential event, distribution historical characteristics are obtained for the second commercial tenant which belongs to the same cluster, so that a pre-trained yield prediction model is expanded and applied to the commercial tenant which does not participate in the preferential event. And predicting yield before distributing the benefits, and screening merchants and distributing the benefits by using the model prediction result and/or the historical yield set, so that the benefit return of the development of the benefits is improved.

In the technical scheme of the disclosure, the relevant data of the related merchants are collected, stored, used, processed, transmitted, provided, published, applied and the like, all meet the regulations of relevant laws and regulations, necessary confidentiality measures are taken, and the customs of the public order is not violated.

Fig. 1 schematically shows an application scenario diagram of a merchant offer distribution method according to an embodiment of the present disclosure.

As shown in fig. 1, the application scenario 100 according to this embodiment may include

terminal devices

101, 102, 103, a network 104 and a server 105. Network 104 is the medium used to provide communication links between

terminal devices

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

A user may use

terminal devices

101, 102, 103 to interact with a server 105 over a network 104 to receive or send messages or the like. The

terminal devices

101, 102, 103 may have installed thereon various communication client applications, such as shopping-like applications, web browser applications, search-like applications, instant messaging tools, mailbox clients, social platform software, etc. (by way of example only).

The

terminal devices

101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, desktop computers, and the like.

The server 105 may be a server providing various services, such as a background management server (for example only) providing support for websites browsed by users using the

terminal devices

101, 102, 103. The background management server may analyze and perform other processing on the received data such as the user request, and feed back a processing result (e.g., a webpage, information, or data obtained or generated according to the user request) to the terminal device.

It should be noted that the merchant offer distribution method provided by the embodiment of the present disclosure may be generally executed by the server 105. Accordingly, the merchant offer distribution apparatus provided by the embodiments of the present disclosure may be generally disposed in the server 105. The merchant offer distribution method provided by the embodiment of the present disclosure may also be executed by a server or a server cluster that is different from the server 105 and is capable of communicating with the

terminal devices

101, 102, 103 and/or the server 105. Correspondingly, the merchant offer distribution device provided by the embodiment of the present disclosure may also be disposed in a server or a server cluster different from the server 105 and capable of communicating with the

terminal devices

101, 102, 103 and/or the server 105.

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

The merchant preferential allocation method according to the embodiment of the present disclosure will be described in detail below with reference to fig. 2 to 7 based on the scenario described in fig. 1.

Fig. 2 schematically shows a flowchart of a merchant offer distribution method according to an embodiment of the present disclosure.

As shown in fig. 2, the merchant offer distribution method of this embodiment includes operations S210 to S240.

In operation S210, merchant feature clustering is performed on N first merchants and M second merchants to obtain S cluster clusters, where the merchant features are obtained according to inherent attribute data of the merchants, the N first merchants include merchants who have participated in historical discount activities, the M second merchants include merchants who have not participated in historical discount activities, and N, M and S are integers greater than or equal to 1, respectively;

illustratively, clustering may be performed using a K-means clustering algorithm, a KNN (K-Nearest Neighbors) clustering algorithm, or a Gaussian mixture clustering algorithm, or the like. The attribute data inherent to the merchant includes data irrelevant to the preferential activities of the merchant in the production and management activities, such as monthly active customer amount of the merchant, monthly transaction amount of the merchant, commission amount of the merchant for receiving the order in the last year or sales type of the merchant. The historical offer activities include activities for offering distribution to merchants that have ended.

In operation S220, obtaining an allocation history feature of each second merchant in the first cluster according to a history feature of at least one first merchant in the first cluster, where the first cluster is any one of the S clusters, and the history feature is obtained according to data of the at least one first merchant in the process of participating in the historical discount activity;

illustratively, all merchants are classified into an S-class by a clustering algorithm. In the s-th class, if there are s1 merchants without participating in historical activities and s2 merchants with participating in historical activities in the first cluster, the average of the sum of the historical features of the s2 merchants is assigned to the s1 merchants without participating in historical activities. In this embodiment, historical feature allocation may be performed on merchants without participating in the historical activities in each cluster, or historical feature allocation may be performed on merchants without participating in the historical activities by screening cluster clusters of high-quality merchants.

In operation S230, inputting the distribution historical characteristics and merchant characteristics of each second merchant into a pre-trained yield prediction model to obtain a predicted yield of each second merchant, where the yield prediction model is obtained by training according to the merchant characteristic set, the historical characteristic set, and the historical yield set of N first merchants;

illustratively, the merchant characteristics of each second merchant are the same dimension as the merchant characteristics of the first merchant. The set of merchant characteristics includes merchant characteristics for each first merchant, the set of historical characteristics includes historical characteristics for each first merchant, and the set of historical profitability includes historical profitability for each first merchant.

Illustratively, the profitability prediction model can be constructed according to a machine learning algorithm, such as random forest, logistic regression, support vector machine, bayesian or neural network, and the like. Taking the neural network model as an example, it may be a classification model, specifically, the profitability is classified into different categories, such as five categories (just examples) from high to low, 20% (class 1 profitability), 40% (class 2 profitability), 60% (class 3 profitability), 80% (class 4 profitability), 100% (class 5 profitability), and the predicted profitability may belong to any one of the five categories.

In operation S240, merchant offer distribution is performed according to the predicted revenue rate sets and/or the historical revenue rate sets of the M second merchants.

Illustratively, the feature vector F (F1 | F2) is also obtained for merchants without participation in the event history, where F1 is a merchant feature and F2 is an allocation history feature. And inputting the characteristic vector F (F1 | F2) as an input into the yield prediction model to obtain the predicted yield of the merchants without participating in the activity, and calculating the corresponding yield of all merchants by combining the yield of the merchants with the history of participating in the activity. In some embodiments, preferential distribution may be performed only for M second merchants, only for N first merchants, or for all merchants according to actual scenarios.

Embodiments of the present disclosure take into account historical attribute characteristics of merchants as compared to making revenue predictions based directly on current merchant characteristics. And on the basis of the first commercial tenant which participates in the historical preferential event, distributing historical characteristics for a second commercial tenant in the same cluster by using a clustering method, and expanding preferential distribution to the commercial tenants which do not participate in the preferential event through a pre-trained profit prediction model. And predicting yield before distributing the benefits, and screening and distributing the benefits by using the model prediction result and/or the historical yield set to obtain a reasonable benefit distribution scheme and promote the benefit return of the development of the benefit activities.

Fig. 3 schematically illustrates a flow diagram of feature screening according to an embodiment of the present disclosure.

Before training the profitability prediction model, as shown in fig. 3, the feature screening of this embodiment includes operations S310 to S330.

In operation S310, obtaining a candidate feature set of each first merchant of the N first merchants, where the candidate feature set includes a candidate merchant feature and a candidate history feature of each first merchant, the candidate merchant feature of each first merchant includes at least one merchant sub-feature, and the candidate history feature of each first merchant includes at least one history sub-feature;

illustratively, for a sample with historical characteristics, multi-dimensional characteristic information of a merchant participating in historical preferential activities is firstly acquired as a candidate characteristic set, and the candidate characteristic set comprises the historical characteristics such as monthly active customer quantity of the merchant, monthly transaction amount of the merchant, past annual income of the merchant and the like, and transaction amount, income amount or preferential categories during the participation of the historical preferential activities. Wherein the candidate merchant features and the candidate historical features of each first merchant are multidimensional and are marked as F (F) ₁ ，f ₂ ...f _x ) Each dimension may be referred to as a merchant sub-feature, such as a merchant sub-feature of three dimensions, namely, the monthly active customer amount of the merchant, the monthly transaction amount of the merchant, and the annual commission return of the merchant. Similarly, each history sub-feature is a dimension of the history feature.

In operation S320, calculating an information gain of each merchant sub-feature and each history sub-feature in the candidate feature set;

and screening a certain number of characteristics from the information gain by using a decision tree.

Fig. 4 schematically illustrates a flow chart for calculating an information gain according to an embodiment of the present disclosure.

As shown in fig. 4, calculating the information gain of this embodiment includes operations S410 to S430. Operation S430 is one embodiment of operation S320.

In operation S410, obtaining a historical revenue record of each first merchant;

in operation S420, classifying the N first merchants according to the historical revenue records;

illustratively, the latest income records of y merchants with historical preferential income can be extracted as historical income records, and the merchants are classified into k categories according to the latest income records. For example, high, medium and low categories according to the amount of the benefit.

In operation S430, an information gain of each merchant sub-feature and each history sub-feature in the candidate feature set is calculated according to the classification results of the N first merchants.

Illustratively, the information gain may be understood as a change in information occurring before and after dividing the data set. The information gain may be obtained by subtracting the conditional entropy from the information entropy. The information gain for each candidate sub-feature (each merchant sub-feature or each historical sub-feature) is as in equation 1).

GAIN(D，f _x )＝H(D)-H(D|f _x ) Formula 1)

Wherein,

wherein p is _i Representing any candidate sub-feature as belonging to the feature f _x Probability of the determined category. D _i Representing features f in a candidate feature set D _x Determined category correspondenceA subset of (a).

By way of example above, merchants are classified into 3 categories (high, medium, low), for example, based on revenue amounts. Based on the feature f _x Subdividing these merchants into classes 3 (e.g., f) _x For the scale of the merchant, the merchant can be considered to be high according to the large scale, medium scale and low scale of the merchant).

According to f _x After the merchants are divided, the 3-type distribution D of one merchant is formed again, which is different from the distribution formed by classifying according to the income amount at first. For example, there are 3 high merchants, 3 medium merchants, 4 low merchants according to the amount before, according to the feature f _x There may be 8 high merchants, 1 medium merchant, 1 low merchant. Then H (Di) represents the entropy under this distribution (8,1,1), which can be calculated by applying the calculation formula for H (D), and then H (D | fx) can be calculated.

In operation S330, each merchant sub-feature and each history sub-feature are filtered according to the information gain by using a decision tree algorithm, so as to obtain a merchant feature set and a history feature set.

Illustratively, a decision tree may be constructed based on an ID3 algorithm according to the information gain result, so as to screen out the finally determined one or more merchant sub-features of each first user to obtain a merchant feature set, and the one or more historical sub-features of each first user to obtain a historical feature set.

In some embodiments, the top n (n ≧ 1) candidate sub-features with the largest information gain (which may be filtered according to the merchant feature and the history feature respectively or may be filtered together) may be selected, and the information gain normalization is used as a weight and multiplied by a vector point containing the top n candidate sub-features to form a feature vector F '(F' ₁ ，f′ ₂ ...f _n ). Obtaining a feature vector F ' (F ') for each first merchant ' ₁ ，f′ ₂ ...f _n ) As a merchant feature set and a historical feature set.

According to the embodiment of the disclosure, the information gain and the decision tree algorithm are utilized to carry out feature screening, so that the conditions that some merchants are limited by the sale types and are not sensitive to preferential activities although some current merchants have considerable features can be avoided.

FIG. 5 schematically illustrates a flow chart for obtaining historical profitability in accordance with an embodiment of the present disclosure.

As shown in fig. 5, obtaining the historical profitability of the embodiment includes operations S510 to S530.

In operation S510, a candidate historical profitability set of each first merchant of the N first merchants is obtained, where the candidate historical profitability set includes at least one sub profitability, and the at least one sub profitability corresponds to at least one historical discount joined by each first merchant one to one;

for example, the rate of return may include a rate of return for an amount of offer allocation by the merchant to participate in the historical offer. For example, if the preferential allocation amount is 10000 Yuan and the profit is 5000 Yuan, the profit rate is 0.5. For example, for each participation in the historical offer during a predetermined period of time (e.g., within a year), the rate of return for that activity may be obtained as a sub-rate of return.

In operation S520, allocating a time decay factor to each sub-rate of return in the at least one sub-rate of return according to a time sequence between the at least one historical offer;

for P-participated in T (T) ₁ ，t ₂ …t _q And T is more than or equal to q and more than or equal to 1) times of historical preferential activities, the performance of each time of participation in the activities can be used as a reference basis for predicting the income of the activities. For the same merchant, the historical performances in different periods have different reference meanings. The reference meaning of the history closer to the present is larger than that of the history in the past, refer to equation 2).

w(t _q )＞w(t _q-1 ) Formula 2)

For example, the first merchant a has engaged in 5 historical offers, and the time decay factor ρ may be assigned according to how far the 5 historical offers are from the suboptimal assignment time. The longer the corresponding historical preferential activity time is, the smaller the corresponding time decay factor is, and the importance of the representation is reduced, refer to formula 3).

ρ _q ＝log ₂ (1-e ^-q ) Formula 3) wherein ρ _g Representing the time decay factor of the q-th order.

In operation S530, a historical rate of return for each first merchant is obtained according to each sub-rate of return and the corresponding time decay factor.

Illustratively, the historical profitability rec of each first merchant may be calculated according to equation 4):

wherein r is _q For the sub-yield of the qth time, | T | is the absolute value of the total historical benefit times of participation.

According to the embodiment of the disclosure, the time decay factor is introduced to process the profitability of the merchants, the importance of each historical preferential activity is evaluated by using the time decay idea, the correlation between the time characteristic of the historical data and the operation performance of the merchants can be considered, and the historical profitability of each merchant can be accurately obtained.

FIG. 6 schematically shows a flow diagram for training a profitability prediction model according to an embodiment of the present disclosure.

Before inputting the distribution history characteristics and the merchant characteristics of each second merchant into the pre-trained yield prediction model, as shown in fig. 6, the method further includes training the yield prediction model, specifically including operations S610 to S630.

In operation S610, inputting N training samples to a yield prediction model to be trained, where the N training samples are obtained according to a merchant feature set, a historical feature set, and a historical yield set of N first merchants, and each training sample includes a merchant feature, a historical feature, and a historical yield of each first merchant;

illustratively, each training sample may be as a vector { F '(F' ₁ ，f′ ₂ ...f _n ) Rec ', where, F ' (F ' ₁ ，f′ ₂ ...f _n ) The training sample is corresponding to the merchant characteristics and historical characteristics of the merchant, rec' the merchant characteristics and historical characteristicsAnd the training sample corresponds to the historical yield category vector of the merchant, if the historical yield obtained according to the formula 4) is less than or equal to 20%, the yield is 1-grade yield, and rec' is (1,0,0,0,0).

In operation S620, the merchant characteristics and the historical characteristics of each first merchant are processed through a profitability prediction model to obtain a predicted profitability of each first merchant;

for example, the profitability prediction model is a classification model constructed based on a neural network, and after merchant features and history features of each first merchant are processed, the profitability category corresponding to the merchant is obtained by using a softmax function, and the predicted profitability is obtained accordingly. For example, the earning rate category of the first merchant a is a class 4 earning rate, the predicted earning rate is determined to be 80%.

In operation S630, parameters of the rate of return prediction model are updated according to a degree of difference between the predicted rate of return and the historical rate of return of each first merchant.

Illustratively, the loss function may be set by cross entropy, and the probability of predicting each first merchant to belong to each rate of return category and rec' get loss function calculation results are input to characterize the degree of difference between the predicted rate of return and the historical rate of return. And according to the gradient descent method, iteratively updating the parameters of the yield prediction model until the loss function converges and the training is finished.

According to the embodiment of the disclosure, the training sample set is obtained by using the commercial tenants participating in the historical activities, and the yield prediction model obtained by training can be applied to commercial tenants not participating in the historical activities to obtain the predicted yield, so that a data base is made for subsequent preferential allocation.

Illustratively, referring to operation S220, the second merchant has merchant characteristics and distribution history characteristics, and thus the predicted profitability of the second merchant can be obtained using the trained profitability prediction model. And combining the historical profitability of the first merchant, and calculating the corresponding profitability by all merchants.

The contents of the merchant offer distribution in operation 240 are described in further detail below.

According to the embodiment of the disclosure, merchant preferential distribution is performed on the N first merchants and the M second merchants according to the predicted profitability of each second merchant and the historical profitability of each first merchant.

In some embodiments, different weights may be given according to the profit rate of each merchant, the business area of the merchant, the amount of passenger or the amount of transaction, and the like, and the evaluation score of each merchant is obtained, and the preferential amount is allocated according to the evaluation score.

In other embodiments, since all merchants have calculated corresponding revenue rates, the total revenue amount may also be obtained from the entirety of each offer, e.g., in combination with the revenue rate and the offer amount. The whole process is converted into a total profit and money optimization problem under the constraint condition of preferential activities.

According to the embodiment of the disclosure, under at least one constraint condition, obtaining the maximum value of the total profit amount by using an optimization algorithm; the at least one constraint includes at least one of: the total preferential allocation quota of the N first commercial tenants and the M second commercial tenants is smaller than or equal to a first preset value; the total preferential allocation quota of each cluster in the S cluster clusters is less than or equal to a corresponding second preset value; the total number of the N first commercial tenants and the M second commercial tenants is less than or equal to a third preset value and greater than or equal to a fourth preset value; the number of the single merchants in each cluster participating in the preferential categories is smaller than or equal to a fifth preset value and larger than or equal to a sixth preset value.

Supposing Y commercial tenants are divided into S classes, and the jth commercial tenant belonging to the ith class is marked as Y _i，j (0 < i < S +1,0 < j < k +1,k is the number of i-th merchant). For each y _i，j And if the second commercial tenant belongs to the second commercial tenant, extracting the commercial tenant feature vector of the second commercial tenant, obtaining distribution historical feature vectors, and calculating the prediction yield of the second commercial tenant according to a yield prediction model. And if the business information belongs to the first merchant, directly obtaining the historical yield of the business information. At this time, assume that the amount of the benefit allocated to each merchant is m _i，j For the above problem, the constraint conditions are as follows, limited by the activity scale and the preferential distribution range.

∑m _i，j ≤M

L ₁ ≤∑NotZero(m _i，j )≤L ₂

Wherein M is the total preferential distribution limit (the first preset value) of the current activity, and M is the total preferential distribution limit _i For the maximum preferential amount (second preset value) allocated to the i-th class (i-th cluster) merchant, L1 and L2 are the lower limit (fourth preset value) and the upper limit (third preset value) of the total merchant number which can participate in the preferential activity, and L _i，1 ，L _i，2 And the lower limit (sixth preset value) and the upper limit (fifth preset value) of the number of the single merchants which can participate in the discount in the ith class are set. The NotZero function is shown below.

According to the embodiment of the disclosure, the merchant preferential distribution to the N first merchants and the M second merchants includes: under the condition that the total income amount of the N first commercial tenants and the M second commercial tenants is maximum, determining preferential allocation amounts of each commercial tenant of the N first commercial tenants and the M second commercial tenants; the total income amount is obtained by calculation according to the income rate of each merchant and the preferential allocation amount, and the income rate of each merchant comprises a predicted income rate or a historical income rate.

Illustratively, the final goal of the merchant offer distribution problem is as shown in equation 5).

E＝max(∑rec _i，j *m _i，j ) Formula 5) wherein E represents the total profit margin, rec _i，j And characterizing the profitability of the jth merchant of the ith class.

Illustratively, the optimization algorithm may include a genetic algorithm, a simulated annealing algorithm, a particle swarm optimization algorithm, an ant colony algorithm, a monte carlo algorithm, or the like.

The process of obtaining the maximum total profit amount will be further described by taking genetic algorithms as examples with reference to steps 1 to 5 below.

Step 1, first generation chromosome generation. Randomly generating a batch of chromosomes as an initial population P = { P = { P = } _a Let its number be v.

And 2, calculating the fitness. For each chromosome P in the starting population P _a Calculating whether the chromosome meets the constraint condition, eliminating the unqualified chromosome, and collecting the qualified chromosomes as P _q ＝{p _bq }。

For each qualified chromosome p _bq Calculating the final profit E _b As shown in equation 6).

E _b ＝rec _i，j *m _i，j Formula 6)

And step 3, naturally selecting. The total yield E of the initial population P is calculated as formula 7).

E＝∑E _b Formula 7)

The probability probabilityb that each chromosome is selected is calculated as shown in equation 8).

Probability _b ＝E _b E type 8)

And 4, generating a new generation of chromosomes. According to probability _b In proportion from P _q Selecting z chromosomes, pairwise matching the chromosomes in the z chromosomes, and pairing the two chromosomes p ₁ And p ₂ In particular, the generation of its progeny p ₃ And p ₄ Is generated randomly at p ₁ Or p ₂ In which the gene (i.e., m) at the half (upper rounded) position is selected _i，j ) And combining with genes at other positions in another chromosome to obtain a new chromosome. The new chromosome combination is P _n 。

Random selection of P _n G chromosomes (g =0.01 × gp, gp being the number of chromosomes in Pn), mutation operations were performed, i.e. the value of a certain gene was randomly changed.

In addition, P is selected _q (v-z) chromosomes with highest medium incomeAnd directly adding the new generation chromosome population Pn into Pn to obtain a new generation chromosome population Pn (if the number is less than v, the number is randomly generated to complement chromosomes).

And 5, iterative reproduction. And (3) repeating the steps 2 to 5 until E meets the requirement (if the variation value is small), ending the iteration, and finally obtaining the chromosome in Pn which is a qualified solution, namely, the optimized merchant preferential allocation scheme can improve the return on the suboptimal allocation.

Fig. 7 schematically illustrates a flow chart of a merchant offer distribution method according to another embodiment of the present disclosure.

As shown in fig. 7, the merchant offer distribution method of this embodiment includes operations S701 to S710.

In operation S701, feature screening is performed. Reference may be made to operations S310 to S330 and operations S410 to S430, which are not described herein.

In operation S702, based on the feature filtering result, merchants having participated in the activity are selected, and a feature vector F (F1 | F' 2) of each merchant is obtained. F1 is a merchant feature, and F'2 is a history feature.

In operation S703, a historical rate of return is determined through time decay. Reference may be made to operations S510 to S530, which are not described herein.

In operation S704, a yield prediction model is trained and obtained using the F (F1 | F' 2) set and the historical yield set of the active merchants. Reference may be made to operations S610 to S630, which are not described herein.

In operation S705, non-participating merchants are selected based on the feature screening result, and a feature vector F (F1) of each merchant is obtained.

In operation S706, for the active and non-active merchants, K-means clustering is performed using F1 of each merchant. Referring to operation S210, details are not described herein.

It should be noted that, in the clustering, all the merchant features before feature screening may also be used.

In operation S707, F2, i.e., an allocation history feature of the non-participating merchant is calculated according to the clustering result. Referring to operation S220, details are not described herein.

In operation S708, the feature vector F (F1 | F2) of each un-participated active merchant is input to the profitability prediction model to obtain the predicted profitability. Referring to operation S230, details are not described herein.

Therefore, historical characteristics of merchants not participating in the activity are supplemented through clustering, the profitability of merchants not participating in the activity is obtained through the model, and the profitability set of all candidate merchants is obtained through integration.

In operation S709, a maximum value of the total profit amount is determined using an optimization algorithm according to the historical profit rate of the merchants having participated in the activity and the predicted profit rate of the merchants having not participated in the activity. Such as optimizing the allocation scheme within constraints through a genetic algorithm. Referring to operation S240, details are not described herein.

In operation S710, an allocation scheme is output. The scheme is that under the condition that the total income amount takes the maximum value, the total income amount comprises the preferential allocation amount obtained by each merchant.

According to the embodiment of the disclosure, compared with a mainstream preference prediction model directly based on the current characteristics, historical performance is added and the trained model is expanded to merchants who do not participate in preference activities. The model is applied to financial institutions such as banks and the like, accurate prediction is carried out before the benefits are distributed, merchant screening and benefit distribution are carried out by using the model prediction result, and the return of carrying out the benefits is promoted.

It should be noted that, although the operations S701 to S710 are described in a sequential order, on the basis of achieving the technical purpose of each step, the execution sequence of each operation is not limited, for example, the operations S702 and S705 may be executed simultaneously, and the operations S704 and S706 may be executed simultaneously.

Based on the merchant discount distribution method, the disclosure also provides a merchant discount distribution device. The apparatus will be described in detail below with reference to fig. 8.

Fig. 8 schematically shows a block diagram of a structure of a merchant offer distribution apparatus according to an embodiment of the present disclosure.

As shown in fig. 8, the merchant benefit distribution apparatus 800 of this embodiment includes a merchant clustering module 810, a feature distribution module 820, a predicted profit rate module 830, and a benefit distribution module 840.

The merchant distance module 810 may perform operation S210, configured to perform merchant feature clustering on N first merchants and M second merchants to obtain S clustered clusters, where the merchant features are obtained according to inherent attribute data of the merchants, the N first merchants include merchants already participating in the historical discount, the M second merchants include merchants not participating in the historical discount, and N, M and S are integers greater than or equal to 1, respectively;

the feature allocating module 820 may perform operation S220, configured to obtain, according to a historical feature of at least one first merchant in a first cluster, an allocated historical feature of each second merchant in the first cluster, where the first cluster is any one of the S clusters, and the historical feature is obtained according to data of the at least one first merchant in the process of participating in the historical discount;

the predicted profitability module 830 may perform operation S230, configured to input the distribution historical characteristics and the merchant characteristics of each second merchant into a pre-trained profitability prediction model to obtain a predicted profitability of each second merchant, where the profitability prediction model is obtained by training according to the merchant characteristic set, the historical characteristic set, and the historical profitability set of the N first merchants;

the offer distribution module 840 may perform operation S240 for distributing the merchant offers according to the predicted revenue rate sets and/or the historical revenue rate sets of the M second merchants.

According to an embodiment of the present disclosure, the merchant offer distribution apparatus 800 may include a feature screening module, which may perform operations S310 to S330 and operations S410 to S430, which are not described herein again.

According to an embodiment of the present disclosure, the merchant offer distribution apparatus 800 may include a time decay module, and the time decay module may perform operations S510 to S530, which are not described herein.

According to an embodiment of the present disclosure, the merchant offer distribution apparatus 800 may include a model training module, which may perform operations S610 to S630, which are not described herein again.

It should be noted that the implementation, solved technical problems, implemented functions, and achieved technical effects of each module/unit/subunit and the like in the apparatus part embodiment are respectively the same as or similar to the implementation, solved technical problems, implemented functions, and achieved technical effects of each corresponding step in the method part embodiment, and are not described herein again.

According to an embodiment of the present disclosure, any of the merchant clustering module 810, the feature allocation module 820, the predicted profitability module 830 and the benefit allocation module 840 may be combined and implemented in one module, or any one of them may be split into a plurality of modules. Alternatively, at least part of the functionality of one or more of these modules may be combined with at least part of the functionality of the other modules and implemented in one module.

According to an embodiment of the disclosure, at least one of the merchant clustering module 810, the feature allocation module 820, the predicted profitability module 830 and the offer allocation module 840 may be implemented at least in part as a hardware circuit, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a system on a chip, a system on a substrate, a system on a package, an Application Specific Integrated Circuit (ASIC), or by hardware or firmware in any other reasonable manner of integrating or packaging a circuit, or by any one of three implementations of software, hardware and firmware, or by a suitable combination of any several of them. Alternatively, at least one of merchant clustering module 810, feature assignment module 820, predicted rate of return module 830, and offer assignment module 840 may be at least partially implemented as a computer program module that, when executed, may perform corresponding functions.

Fig. 9 schematically illustrates a block diagram of an electronic device suitable for implementing a merchant offer distribution method according to an embodiment of the disclosure.

As shown in fig. 9, an electronic apparatus 900 according to an embodiment of the present disclosure includes a processor 901 which can perform various appropriate actions and processes in accordance with a program stored in a Read Only Memory (ROM) 902 or a program loaded from a storage portion 908 into a Random Access Memory (RAM) 903. Processor 901 may comprise, for example, a general purpose microprocessor (e.g., a CPU), an instruction set processor and/or associated chipset, and/or a special purpose microprocessor (e.g., an Application Specific Integrated Circuit (ASIC)), among others. The processor 901 may also include on-board memory for caching purposes. The processor 901 may comprise a single processing unit or a plurality of processing units for performing the different actions of the method flows according to embodiments of the present disclosure.

In the RAM 903, various programs and data necessary for the operation of the electronic apparatus 900 are stored. The processor 901, ROM 902, and RAM 903 are connected to each other by a bus 904. The processor 901 performs various operations of the method flows according to the embodiments of the present disclosure by executing programs in the ROM 902 and/or the RAM 903. Note that the program may also be stored in one or more memories other than the ROM 902 and the RAM 903. The processor 901 may also perform various operations of the method flows according to embodiments of the present disclosure by executing programs stored in the one or more memories.

Electronic device 900 may also include input/output (I/O) interface 905, input/output (I/O) interface 905 also connected to bus 904, according to an embodiment of the present disclosure. The electronic device 900 may also include one or more of the following components connected to the I/O interface 905: an input portion 906 including a keyboard, a mouse, and the like; an output section 907 including components such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage portion 908 including a hard disk and the like; and a communication section 909 including a network interface card such as a LAN card, a modem, or the like. The communication section 909 performs communication processing via a network such as the internet. The drive 910 is also connected to the I/O interface 905 as necessary. A removable medium 911 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 910 as necessary so that a computer program read out therefrom is mounted into the storage section 908 as necessary.

The present disclosure also provides a computer-readable storage medium, which may be contained in the apparatus/device/system described in the above embodiments; or may exist alone without being assembled into the device/apparatus/system. The computer-readable storage medium carries one or more programs which, when executed, implement the method according to an embodiment of the disclosure.

According to embodiments of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example but is not limited to: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present disclosure, a computer 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. For example, according to embodiments of the present disclosure, a computer-readable storage medium may include the ROM 902 and/or the RAM 903 described above and/or one or more memories other than the ROM 902 and the RAM 903.

Embodiments of the present disclosure also include a computer program product comprising a computer program containing program code for performing the method illustrated in the flow chart. When the computer program product runs in a computer system, the program code is used for causing the computer system to realize the method provided by the embodiment of the disclosure.

The computer program performs the above-described functions defined in the system/apparatus of the embodiments of the present disclosure when executed by the processor 901. The systems, apparatuses, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the present disclosure.

In one embodiment, the computer program may be hosted on a tangible storage medium such as an optical storage device, a magnetic storage device, or the like. In another embodiment, the computer program may also be transmitted in the form of a signal over a network medium, distributed, and downloaded and installed via the communication section 909 and/or installed from the removable medium 911. The computer program containing program code may be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.

In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 909, and/or installed from the removable medium 911. The computer program, when executed by the processor 901, performs the above-described functions defined in the system of the embodiment of the present disclosure. The systems, devices, apparatuses, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the present disclosure.

In accordance with embodiments of the present disclosure, program code for executing computer programs provided by embodiments of the present disclosure may be written in any combination of one or more programming languages, and in particular, these computer programs may be implemented using high level procedural and/or object oriented programming languages, and/or assembly/machine languages. The programming language includes, but is not limited to, programming languages such as Java, C + +, python, the "C" language, or the like. The program code may execute entirely on the user computing device, partly on the user device, partly on a remote computing device, or entirely on the remote computing device or server. In situations involving remote computing devices, the remote computing devices 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 external computing devices (e.g., through the internet using an internet service provider).

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

It will be appreciated by a person skilled in the art that various combinations or/and combinations of features recited in the various embodiments of the disclosure and/or in the claims may be made, even if such combinations or combinations are not explicitly recited in the disclosure. In particular, various combinations and/or combinations of the features recited in the various embodiments and/or claims of the present disclosure may be made without departing from the spirit or teaching of the present disclosure. All such combinations and/or associations are within the scope of the present disclosure.

The embodiments of the present disclosure have been described above. However, these examples are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although the embodiments are described separately above, this does not mean that the measures in the embodiments cannot be used advantageously in combination. The scope of the disclosure is defined by the appended claims and equivalents thereof. Various alternatives and modifications can be devised by those skilled in the art without departing from the scope of the present disclosure, and such alternatives and modifications are intended to be within the scope of the present disclosure.

Claims

1. A merchant offer distribution method comprises the following steps:

carrying out merchant characteristic clustering on N first merchants and M second merchants to obtain S clustering clusters, wherein merchant characteristics are obtained according to inherent attribute data of the merchants, the N first merchants comprise merchants participating in historical preferential activities, the M second merchants comprise merchants not participating in the historical preferential activities, and N, M and S are integers greater than or equal to 1 respectively;

obtaining distribution historical characteristics of each second merchant in a first cluster according to historical characteristics of at least one first merchant in the first cluster, wherein the first cluster is any one of the S clusters, and the historical characteristics are obtained according to data of the at least one first merchant in the process of participating in the historical preferential event;

inputting the distribution historical characteristics and the merchant characteristics of each second merchant into a pre-trained yield prediction model to obtain the predicted yield of each second merchant, wherein the yield prediction model is obtained by training according to the merchant characteristic set, the historical characteristic set and the historical yield set of the N first merchants;

and carrying out merchant preferential distribution according to the predicted income rate set and/or the historical income rate set of the M second merchants.

2. The method of claim 1, wherein prior to training the rate of return prediction model, comprising:

obtaining a candidate historical rate of return set of each first merchant of the N first merchants, wherein the candidate historical rate of return set comprises at least one sub rate of return, and the at least one sub rate of return corresponds to at least one historical discount joined by each first merchant one by one;

distributing a time attenuation factor to each sub-rate of return in the at least one sub-rate of return according to the time sequence among the at least one historical preferential activity;

and obtaining the historical profitability of each first merchant according to each sub profitability and the corresponding time attenuation factor.

3. The method of claim 1, wherein prior to training the rate of return prediction model, comprising:

acquiring a candidate feature set of each first merchant in the N first merchants, wherein the candidate feature set includes a candidate merchant feature and a candidate history feature of each first merchant, the candidate merchant feature of each first merchant includes at least one merchant sub-feature, and the candidate history feature of each first merchant includes at least one history sub-feature;

calculating the information gain of each merchant sub-feature and each history sub-feature in the candidate feature set;

and screening each merchant sub-feature and each history sub-feature according to the information gain by using a decision tree algorithm to obtain the merchant feature set and the history feature set.

4. The method of claim 3, comprising:

acquiring a historical income record of each first merchant, wherein the historical income record comprises income obtained by each first merchant participating in the historical preferential event;

classifying the N first commercial tenants according to the historical income records;

wherein the calculating the information gain of each merchant sub-feature and each historical sub-feature in the candidate feature set comprises:

and calculating the information gain of each merchant sub-feature and each history sub-feature in the candidate feature set according to the classification results of the N first merchants.

5. The method according to any one of claims 2 to 4, wherein before inputting the allocation history characteristics and merchant characteristics of each second merchant into a pre-trained rate of return prediction model, training the rate of return prediction model further comprises:

inputting N training samples into the yield prediction model to be trained, wherein the N training samples are obtained according to a merchant feature set, a historical feature set and a historical yield set of the N first merchants, and each training sample comprises a merchant feature, a historical feature and a historical yield of each first merchant;

processing the merchant characteristics and the historical characteristics of each first merchant through the yield prediction model to obtain the predicted yield of each first merchant;

and updating the parameters of the yield prediction model according to the difference degree between the predicted yield and the historical yield of each first merchant.

6. The method of claim 2, wherein the distributing merchant offers according to the set of predicted rates of return and/or the set of historical rates of return for the M second merchants comprises:

and performing merchant preferential distribution on the N first merchants and the M second merchants according to the predicted yield rate of each second merchant and the historical yield rate of each first merchant.

7. The method of claim 6, wherein the merchant offer distribution to the N first merchants and the M second merchants comprises:

under the condition that the total income amount of the N first merchants and the M second merchants is maximum, determining preferential allocation amount of each merchant of the N first merchants and the M second merchants;

and calculating the total income amount according to the income rate and the preferential allocation amount of each merchant, wherein the income rate of each merchant comprises a predicted income rate or a historical income rate.

8. The method of claim 7, further comprising:

under at least one constraint condition, obtaining the maximum value of the total income amount by using an optimization algorithm;

wherein the at least one constraint includes at least one of:

the total preferential distribution amount of the N first commercial tenants and the M second commercial tenants is smaller than or equal to a first preset value;

the total preferential distribution amount of each cluster in the S cluster clusters is smaller than or equal to a corresponding second preset value;

the number of total merchants in the N first merchants and the M second merchants is less than or equal to a third preset value and greater than or equal to a fourth preset value;

the number of the single merchants in each cluster participating in the preferential categories is smaller than or equal to a fifth preset value and larger than or equal to a sixth preset value.

9. A merchant offer distribution apparatus comprising:

the merchant distance module is used for carrying out merchant characteristic clustering on N first merchants and M second merchants to obtain S clustering clusters, wherein merchant characteristics are obtained according to inherent attribute data of the merchants, the N first merchants comprise merchants participating in historical discount activities, the M second merchants comprise merchants not participating in the historical discount activities, and N, M and S are integers larger than or equal to 1 respectively;

the characteristic distribution module is used for obtaining distribution historical characteristics of each second commercial tenant in a first cluster according to historical characteristics of at least one first commercial tenant in the first cluster, wherein the first cluster is any one of the S cluster, and the historical characteristics are obtained according to data of the at least one first commercial tenant in the historical preferential activity process;

the prediction yield rate module is used for inputting the distribution historical characteristics and the merchant characteristics of each second merchant into a pre-trained yield rate prediction model to obtain the prediction yield rate of each second merchant, wherein the yield rate prediction model is obtained by training according to the merchant characteristic set, the historical characteristic set and the historical yield rate set of the N first merchants;

and the discount distribution module is used for carrying out merchant discount distribution according to the predicted income rate set and/or the historical income rate set of the M second merchants.

10. An electronic device, comprising:

one or more processors;

a storage device 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 perform the method of any of claims 1-8.

11. A computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to perform the method according to any one of claims 1 to 8.

12. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 8.