Transaction data processing method and transaction data processing system

Abstract

Provided are a transaction data processing method and system. The method comprises: acquiring a target proportion vector of a target across different item categories, and constructing a target proportion vector model sample on the basis of the target proportion vector; constructing an association model on the basis of the target proportion vector model sample and an association probability of the target being associated from one item category to another item category; constructing a transition model on the basis of the target proportion vector model sample, the association probability, and a transition association probability of the target being transitioned from one item category to another item category; performing parameter fitting on parameters of the transition model on the basis of the target proportion vector model sample to obtain the transition association probability; and computing target-related decisions for the different item categories on the basis of the transition association probability. Therefore, changes in proportions of a target in different item categories caused by participation of an acting object can be efficiently and accurately identified, and the accurate identification of the changes in proportion allows for optimization of resource allocation and supports effective risk management.

Description

Transaction data processing method and transaction data processing system

[0001] The present application claims priority to Chinese Patent Application No. 2024111336823, filed on August 16, 2024, and entitled "Transaction data processing method and transaction data processing system", the entire disclosure of which is incorporated herein by reference. TECHNICAL FIELD

[0002] The present application relates to the field of computer information technology, and particularly relates to a transaction data processing method and a transaction data processing system. BACKGROUND

[0003] With the development of computer internet technology, the types and ways of transactions are also increasing. In order to stimulate consumption, governments, financial institutions or enterprises will issue resources with certain rights and interests, such as consumption coupons. Users can enjoy the rights and interests of the resources by obtaining the resources and transacting and canceling at designated places, which can stimulate user consumption to a certain extent.

[0004] However, the funds for issuing consumption coupons are limited, and the effects of issuing consumption coupons in different industries at different proportions are uneven. Therefore, how to allocate consumption coupons, especially in which industries to allocate, and at what proportion to allocate, in order to optimize the effect of consumption coupon allocation, is also a key technical problem to be solved.

[0005] On the other hand, the existing methods for consumption coupon allocation mainly target a single customer group, which is difficult to meet the allocation needs of the government facing industries. Determining the target industries and the allocation of the amount of issuance in each industry by using artificial methods for related technologies also leads to the problem of poor effect of consumption coupon issuance. Currently, there is no effective solution. SUMMARY

[0006] In order to solve the above-mentioned problems in the prior art, the present application aims to provide a transaction data processing method and a transaction data processing system which can efficiently and accurately find out the changes in the target proportion of the target in different project categories caused by the participation of the action object.

[0007] Further, the present application aims to provide a transaction data processing method and a transaction data processing system which can efficiently and accurately allocate resources with rights and interests (such as consumption coupons).

[0008] The transaction data processing method of one aspect of the present application comprises:

[0009] A model sample construction step, obtaining a target proportion vector of a target in different project categories, and constructing a target proportion vector model sample based on the target proportion vector;

[0010] a correlation model construction step of constructing a correlation model based on the target proportion vector model sample and a correlation probability of a target being correlated from one item category to another item category;

[0011] a transition model construction step of constructing a transition model based on the target proportion vector model sample, the correlation probability, and a transition correlation probability of a target being transitioned from one item category to another item category;

[0012] a parameter fitting step of fitting parameters of the transition model based on the target proportion vector model sample to obtain the transition correlation probability; and

[0013] a decision calculation step of calculating a decision related to a target for different item categories based on the transition correlation probability.

[0014] Optionally, in the transition model construction step, the transition model is constructed based on the target proportion vector model sample, the correlation probability, the transition correlation probability, and an action object identifier for identifying whether to participate in a transaction activity.

[0015] Optionally, in the correlation model construction step, the correlation model is constructed based on a Markov process,

[0016] In the transition model construction step, an action sample participating in a transaction activity and a non-action sample not participating in a transaction activity are included in the Markov process to construct the transition model.

[0017] Optionally, in the parameter fitting step, any one of the following fitting models is used for parameter fitting:

[0018] a least square estimation fitting model, a maximum likelihood estimation fitting model, a moment estimation fitting model, a least absolute deviation estimation fitting model, and a generalized moment estimation fitting model.

[0019] Optionally, when there are P different item categories, the correlation probability is represented by a P×P dimensional correlation matrix, and the transition correlation probability is represented by a P×P dimensional correlation matrix, where P is a natural number.

[0020] Optionally, in the decision calculation step, as the decision related to the target, a sum of each column of the P columns of the transition correlation probability is calculated and sorted to obtain a priority order.

[0021] Optionally, in the decision calculation step, as the decision related to the target, n item categories are selected from the P categories, and a proportion of each of the n item categories is calculated.

[0022] Optionally, in the model sample construction step, a consumption proportion vector of the consumer in different industry categories is obtained as the target proportion vector, and a model sample is constructed according to the following formula (1), wherein it is assumed that there are P different industries:

[0023] Y it = (y it,1 , y it,2 ,..., y it,P ) T (1)

[0024] wherein Y it is the consumption proportion vector, i represents the ith sample, t represents the tth time, t-1 represents the t-1th time,

[0025] In the correlation model construction step, a correlation model is constructed based on a Markov process according to the following formula (2):

[0026] Y it = WY it-1 + ε it (2)

[0027] wherein W = (w p1p2 ) ∈ R P×P is the correlation probability, w p1p2 represents a transition probability of a consumption tendency of the consumer in an industry p2 to an industry p1, and ε it is a noise term,

[0028] In the variation model construction step, a transition probability variation model is constructed according to the following formula (3):

[0029] Y it = WY it-1 + Treat i * ΩY it-1 + ε it (3)

[0030] wherein Ω = (τ p1p2 ) ∈ R P×P is the probability variation matrix, τ p1p2 represents a transition probability of a consumption variation of the consumer in an industry p2 to an industry p1, ε it is a noise term, and Treat i represents the action object identifier.

[0031] Optionally, in the correlation model construction step, W is trained such that an error of WY it-1 + ε it and Y it is minimized.

[0032] Optionally, in the decision calculation step, column sum is performed on the probability variation matrix and ranking is performed, and the obtained ranking is taken as the priority ranking of the consumption coupon issuing industry.

[0033] Optionally, in the decision calculation step, n industries are selected to calculate the proportion of the n industries and taken as the proportion of the consumption coupon issuing industry.

[0034] The transaction data processing system of one aspect of the present application comprises:

[0035] A model sample construction module is configured to obtain a target proportion vector of a target in different project categories, and construct a target proportion vector model sample based on the target proportion vector.

[0036] An association model construction module is configured to construct an association model based on the target proportion vector model sample and an association probability of the target from one project category to another project category.

[0037] A variation model construction module is configured to construct a variation model based on the target proportion vector model sample, the association probability, and a variation association probability of the target from one project category to another project category.

[0038] A parameter fitting module is configured to perform parameter fitting on parameters of the variation model based on the target proportion vector model sample to obtain the variation association probability.

[0039] A decision calculation module is configured to calculate a decision related to the target of different project categories based on the variation association probability.

[0040] Optionally, in the variation model construction module, the variation model is constructed based on the target proportion vector model sample, the association probability, the variation association probability, and an action object identifier for identifying whether to participate in a transaction activity.

[0041] In the variation model construction step, an action sample participating in a transaction activity and a non-action sample not participating in a transaction activity are included in a Markov process to construct the variation model.

[0042] Optionally, in the association model construction module, the association model is constructed based on a Markov process.

[0043] Optionally, in the parameter fitting module, any one of the following fitting models is used for parameter fitting:

[0044] A least square estimation fitting model, a maximum likelihood estimation fitting model, a moment estimation fitting model, a least one estimation fitting model, and a generalized moment estimation fitting model.

[0045] Optionally, when there are P different categories of items, the association probability is represented by a P×P dimensional association matrix, and the variable association probability is represented by a P×P dimensional association matrix, where P is a natural number.

[0046] Optionally, in the decision calculation module, as the decision related to the target, the sum of each column of the P columns of the variable association probability is calculated and sorted to obtain a priority ranking.

[0047] Optionally, in the decision calculation module, as the decision related to the target, n category items are selected from the P categories, and the proportion of the n category items is calculated respectively.

[0048] Optionally, in the model sample construction module, a consumption proportion vector of a consumer in different industry categories is obtained as the target proportion vector, and a model sample is constructed according to the following formula (1), where it is assumed that there are P different industries:

[0049] Y it = (y it,1 , y it,2 ,..., y it,P ) T (1)

[0050] where Y it is the consumption proportion vector, i represents the i-th sample, t represents the t-th time, and t-1 represents the t-1-th time.

[0051] In the association model construction module, an association model is constructed based on a Markov process according to the following formula (2):

[0052] Y it = WY it-1 + ε it (2)

[0053] where W = (w p1p2 ) ∈ R P×P is the association probability, w p1p2 represents the transition probability of the consumer's consumption inclination in industry p2 to industry p1, and ε it is a noise term.

[0054] In the variable model construction module, a transition probability variable model is constructed according to the following formula (3):

[0055] Y it = WY it-1 + Treat i * ΩY it-1 + ε it (3)

[0056] where Ω = (τp1p2 )∈R P×P is the probability variation matrix, τ p1p2 represents the transition probability of the consumer's consumption variation in industry p2 to industry p1, ε it is a noise term, Treat i represents the action object identifier.

[0057] Optionally, in the association model construction module, W is trained such that the error of WY it-1 +ε it and Y it is minimized.

[0058] Optionally, in the decision calculation module, column summation is performed on the probability variation matrix and sorting is performed, and the priority ranking obtained is taken as the ranking of the industries to which the consumer coupons are put.

[0059] Optionally, in the decision calculation module, n industries are selected to calculate the proportion of the n industries and take the proportion as the proportion of the industries to which the consumer coupons are put.

[0060] The computer readable medium of one aspect of the application has a computer program stored thereon, and the computer program is executed by a processor to implement the transaction data processing method described above.

[0061] The computer device of one aspect of the application comprises a storage module, a processor, and a computer program stored on the storage module and executable on the processor, and the processor implements the transaction data processing method described above when executing the computer program.

[0062] The computer program product of one aspect of the application comprises a computer program, and the computer program is executed by a processor to implement the transaction data processing method described above. BRIEF DESCRIPTION OF DRAWINGS

[0063] FIG. 1 is a flow diagram of the transaction data processing method of the application.

[0064] FIG. 2 is a structural block diagram of the transaction data processing system of the application.

[0065] The above and other objects and advantages of the present application will become more apparent from the following detailed description taken in conjunction with the accompanying drawings, in which like elements are denoted by like reference numerals, and in which: DETAILED DESCRIPTION

[0066] The following describes some of the multiple embodiments of the application, which are intended to provide a basic understanding of the application. It is not intended to identify key or decisive elements or to limit the scope of protection.

[0067] For the sake of brevity and clarity, the principles of the application are described herein primarily in terms of an exemplary implementation with reference to drawings. One skilled in the art, however, will readily recognize that the same principles apply to and can be implemented in all types of transaction data processing methods and transaction data processing systems, and that any such variations are within the true spirit and scope of the present patent application.

[0068] Moreover, in the following description, reference is made to the accompanying drawings that form a part hereof, and in which are shown, by way of illustration, specific implementations. Changes in the electrical, mechanical, logical, and structural

[0069] The terms such as "have" and "include" mean that there are no other elements (modules) and steps in addition to the elements (modules) and steps directly and explicitly expressed in the specification and claims, and the technical solutions of the present application do not exclude the presence of other elements (modules) and steps not directly or explicitly expressed.

[0070] The transaction data processing method and the transaction data processing system of the present application can not only be applied to the distribution of consumer coupons, but also be applied to risk prevention and control, e-commerce and various aspects. In the following description, consumer coupons are taken as examples for the convenience of illustration.

[0071] Figure 1 is a flowchart of the transaction data processing method of the present application.

[0072] As shown in Figure 1, the transaction data processing method of the present application comprises the following steps:

[0073] Model sample construction step S10): Obtain the consumption proportion vector Y of consumers in different industry categories it , and construct a model sample:

[0074] Suppose there are P different industries, Y it is a P-dimensional vector:

[0075] Y it =(y it,1 , y it,2 ,..., y it,P ) T ,

[0076] Where i represents the i-th sample, and t represents the t-th time;

[0077] Correlation model construction step S200: industry consumption correlation model construction based on Markov process:

[0078] Y it = WY it-1 + ε it ,

[0079] wherein W = (w p1p2 ) e R P×P is a P x P dimensional transition matrix, w p1p2 represents the transition probability of the consumer's consumption inclination in industry p2 to industry p1, and ε it is a noise term;

[0080] Variation model construction step S300: fusing the identification of whether to participate in the activity into the fitting process of the model to construct a transition probability variation model:

[0081] Y it = WY it-1 + Treat i * Omega Y it-1 + epsilon it ,

[0082] wherein Omega = (tau p1p2 ) e R P×P is a P x P dimensional probability variation matrix, tau p1p2 represents the transition probability of the consumer's consumption variation in industry p2 to industry p1, epsilon it is a noise term, and Treat i represents the identification of whether the sample is the action object of the activity, which is an important point of the present application, that is, the identification of whether to participate in the activity is fused into the fitting process of the model;

[0083] Parameter fitting step S400: fitting the parameters of the transition probability variation model constructed in the transition probability variation model construction step S300 based on the sample constructed in the sample construction step S100, wherein the parameter estimation model includes but is not limited to an OLS least square estimation fitting model; and

[0084] Decision calculation step S500: calculating the priority ranking and the allocation ratio value of the consumption coupon allocation decision, specifically, based on the probability variation matrix Omega fitted in the parameter fitting step S400, calculating the sum of each column of the matrix and ranking to obtain the priority ranking of the consumption coupon allocation, and selecting Top n (wherein n is determined according to the number of industries that can be taken), and calculating the allocation ratio of the n industries.

[0085] According to the transaction data processing method of the present application, the change in the proportion of the target in different project categories caused by the participation of the action object can be efficiently and accurately found out, and the change in the proportion can be accurately found out to optimize resource allocation and support effective risk management.

[0086] Next, one embodiment is taken as an example to be described below, in which the issuance of consumption coupons in a certain region is taken as an example for description.

[0087] The transaction data processing method of this embodiment comprises the following steps:

[0088] (1) Obtain the consumption proportion vector Y of consumers in different industry categories based on the index statistical method it , construct a model training sample, and assign a label of whether to participate in the consumption coupon activity, 1 for participation, and 0 otherwise. For ease of description, three industry categories, i.e., catering, retail, and electronic products, are taken as examples for description, for example, the consumption proportion vector Y of the i-th user in the catering, retail, and electronic product industries at the t-th moment is Y it =[0.3, 0.2, 0.5], and the consumption proportion vector Y of the i-th user in the catering, retail, and electronic product industries at the t+1-th moment is Y it+1 =[0.2, 0.2, 0.6].

[0089] (2) The industry consumption correlation model based on Markov process is constructed as follows: Y it =WY it-1 +ε it ,

[0090] Wherein, Markov process is an important class of random processes, its main feature is "memoryless" or "Markov property", specifically, if the future state of a system depends only on the current state, and has nothing to do with the past state, then the system or process has Markov property, which can be called Markov process. Markov process is memoryless (future state only depends on current state, and has nothing to do with past state), Markov process can describe transition probability (describe the probability of transition from one state to another state). The present application is realized by using the characteristic that Markov process can describe transition probability.

[0091] Wherein, W=(w p1p2 )∈R 3×3 is a 3x3 dimensional correlation matrix, w p1p2 represents the correlation probability of the consumer's consumption in industry p2 to industry p1, and ε it is a noise term. Here, W is a parameter to be fitted, and ε itis a random number term, including but not limited to a normal distribution, and has a normal distribution characteristic with a mean of 0 and a variance of σ 2 Here, Y it-1 and Y it are inputs of the model, and W is trained to minimize the error of WY it-1 + ε it and Y it . The fitting process will be implemented in (5), and W reflects the consumption correlation between industries.

[0092] (3) The construction of the transition probability variation model of the fusion participation activity identifier is implemented as follows:

[0093] Y it = WY it-1 + Treat i * ΩY it-1 + ε it

[0094] Where, Ω = (τ p1p2 ) ∈ R 3×3 is a 3x3 dimensional probability variation matrix, τ p1p2 represents the correlation probability of consumer consumption in industry p2 to industry p1, ε it is a noise term, and Treat i represents whether the sample is the object of the intervention, i.e., whether it participates in the transaction activity, and is marked as 1 if it has been affected and as 0 if it has not been affected. Here, Ω is the parameter to be fitted, and ε_it is a random number term, including but not limited to a normal distribution, and has a normal distribution characteristic with a mean of 0 and a variance of σ 2 Here, relative to (3), the participation identifier of the activity is integrated, so Ω can better reflect the changes in consumption correlation between industries caused by the consumption coupon.

[0095] Here, by incorporating multiple types of samples (i.e., both samples participating in the transaction activity and non-affected samples not participating in the transaction activity) into the Markov fitting process, the dynamic changes in consumer behavior can be more accurately revealed. In this way, by incorporating the Markov fitting idea of multiple types of samples, not only the accuracy and comprehensiveness of parameter fitting are improved, but also strong support is provided for in-depth understanding of consumer behavior and its impact on activities.

[0096] (4) The parameter fitting of the transition probability variation model of the fusion participation activity identifier is implemented, specifically including:

[0097] Based on the sample constructed in (1), the parameters of the fusion intervention transition probability variation model constructed in (3) are fitted, and the parameter estimation model includes but is not limited to the OLS least squares estimation fitting model. For example: Yit to Y it+1 fitting, [0.2, 0.2, 0.6]≈W[0.3, 0.2, 0.5]+1*Ω[0.3, 0.2, 0.5]+0.25, W and Ω are a 3*3 dimensional matrix parameters.

[0098] (5) Decision analysis based on probability change (substitution effect) mining matrix

[0099] Based on the probability change matrix obtained by fitting in (4)

[0100] As an example, assuming that Top2 industries (i.e. the top two industries) are needed to be recommended, the matrix is column-added and recommended sorting can be obtained [1.00, 1.39, 1.14] -> [1.39, 1.14, 1.00] -> [retail, electronic products, catering] -> [retail, electronic products], and the retail investment proportion is:

[0101] As described above, according to the transaction data processing method of the embodiment, the model automatically fits the consumption coupon investment into an inter-industry consumption association matrix, and the priority ranking and investment proportion value of the output consumption coupon investment are calculated based on the association matrix, thereby solving the problem that the related art determines the target industry and the distribution of the investment amount of each industry in a manual manner, resulting in poor consumption coupon investment effect. Specifically, the transaction data processing method according to the embodiment provides efficient and accurate support for actual consumption coupon investment decision-making. Under the same consumption coupon investment amount, compared with the uniform investment method, the transaction data processing method according to the embodiment improves the overall consumption transaction volume by 4.5%.

[0102] Moreover, according to the transaction data processing method of the embodiment, by constructing a probability change matrix Ω between industries based on a Markov process, the characteristics of the change value are fused, and the fitting effect of the model can be further improved.

[0103] Further, the transaction data processing method of the present application can not only be applied to the field of consumption coupon investment, but also be applied to risk prevention and control, e-commerce and other aspects, and has good application effect and wide application range, and has strong universality.

[0104] FIG. 2 is a structural block diagram of the transaction data processing system of the present application.

[0105] As shown in FIG. 2, the transaction data processing system 100 of the present application comprises:

[0106] The model sample construction module 110 is configured to acquire a target proportion vector of a target in different project categories, and construct a target proportion vector model sample based on the target proportion vector.

[0107] The association model construction module 120 is configured to construct an association model based on the target proportion vector model sample and an association probability of the target from one project category to another project category.

[0108] The variation model construction module 130 is configured to construct a variation model based on the target proportion vector model sample, the association probability, and a variation association probability of the target from one project category to another project category.

[0109] The parameter fitting module 140 is configured to perform parameter fitting on the variation model based on the target proportion vector model sample to obtain the variation association probability.

[0110] The decision calculation module 150 is configured to calculate a decision related to the target in different project categories based on the variation association probability.

[0111] In the variation model construction module 130, the variation model is constructed based on the target proportion vector model sample, the association probability, the variation association probability, and an object identifier for identifying whether to participate in a transaction activity.

[0112] In the association model construction module 120, the association model is constructed based on a Markov process.

[0113] In the parameter fitting module 140, the parameter fitting is performed by using any one of the following fitting models:

[0114] a least square estimation fitting model, a maximum likelihood estimation fitting model, a moment estimation fitting model, a least absolute deviation estimation fitting model, and a generalized moment estimation fitting model.

[0115] When there are P different categories of projects, the association probability is represented by a P×P-dimensional association matrix, and the variation association probability is represented by a P×P-dimensional association matrix, where P is a natural number.

[0116] In the decision calculation module 150, as the decision related to the target, a sum of each column of the P columns of the variation association probability is calculated and sorted to obtain a priority order.

[0117] In the decision calculation module 150, as the decision related to the target, n category projects are selected from the P categories, and a proportion of each of the n category projects is calculated.

[0118] As an embodiment of a consumption coupon issuing application, in the model sample construction module 110, a consumption proportion vector of a consumer in different industry categories is obtained as the target proportion vector, and a model sample is constructed according to the following formula (1), wherein it is assumed that there are P different industries:

[0119] Y it = (y it,1 , y it,2 ,..., y it,P ) T (1)

[0120] wherein Y it is the consumption proportion vector, i represents the i th sample, t represents the t th moment, t-1 represents the t-1 th moment,

[0121] In the association model construction module 120, an association model is constructed based on a Markov process according to the following formula (2):

[0122] Y it = WY it-1 + ε it (2)

[0123] wherein W = (w p1p2 ) ∈ R P×P is the association probability, w p1p2 represents a transition probability of a consumption tendency of a consumer in an industry p2 to an industry p1, and ε it is a noise term,

[0124] In the variation model construction module 130, a transition probability variation model is constructed according to the following formula (3):

[0125] Y it = WY it-1 + Treat i * ΩY it-1 + ε it (3)

[0126] wherein Ω = (τ p1p2 ) ∈ R P×P is the probability variation matrix, τ p1p2 represents a transition probability of a consumption variation of a consumer in an industry p2 to an industry p1, ε it is a noise term, and Treat i represents the action object identifier.

[0127] In the association model construction module 120, W is trained such that the error of WY it-1 + ε it and Y it is minimized.

[0128] In the decision calculation module 150, column summation is performed on the probability variation matrix, and the obtained priority ranking is taken as the ranking of the consumption coupon issuing industry, or n industries are selected to calculate the proportion of the n industries and the proportion of the consumption coupon issuing industry.

[0129] The application further provides a computer readable medium, which stores a computer program, and the computer program is executed by a processor to implement the transaction data processing method.

[0130] The application further provides a computer device, which comprises a storage module, a processor and a computer program stored in the storage module and capable of running on the processor, and the processor implements the transaction data processing method when executing the computer program.

[0131] The application further provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the transaction data processing method.

[0132] The above merely describes specific embodiments of the present application, but the protection scope of the present application is not limited to this. Those skilled in the art can think of other feasible changes or replacements according to the technical range disclosed in the present application, and such changes or replacements are all covered in the protection scope of the present application. The embodiments and features in the embodiments can be combined with each other without conflict. The protection scope of the present application is defined by the claims.

Claims

1. A transaction data processing method characterized by, Comprise: a model sample construction step of obtaining a target proportion vector of a target in different project categories and constructing a target proportion vector model sample based on the target proportion vector; an association model construction step of constructing an association model based on the target proportion vector model sample and an association probability of the target from one project category to another project category; a variation model construction step of constructing a variation model based on the target proportion vector model sample, the association probability and a variation association probability of the target from one project category to another project category; a parameter fitting step of fitting parameters of the variation model based on the target proportion vector model sample to obtain the variation association probability; and a decision calculation step of calculating a decision related to the target in different project categories based on the variation association probability.

2. The transaction data processing method according to claim 1, wherein in the variation model construction step, the variation model is constructed based on the target proportion vector model sample, the association probability, the variation association probability and an action object identifier for identifying whether to participate in a transaction activity.

3. The transaction data processing method according to claim 2, wherein in the association model construction step, the association model is constructed based on a Markov process, in the variation model construction step, the action sample participating in the transaction activity and the non-action sample not participating in the transaction activity are included in the Markov process to construct the variation model.

4. The transaction data processing method according to claim 3, wherein in the parameter fitting step, any one of the following fitting models is used for parameter fitting: a least square estimation fitting model, a maximum likelihood estimation fitting model, a moment estimation fitting model, a least absolute deviation estimation fitting model and a generalized moment estimation fitting model.

5. The transaction data processing method according to claim 4, wherein when there are P different categories of projects, the association probability is represented by a P×P dimensional association matrix and the variation association probability is represented by a P×P dimensional association matrix, wherein P is a natural number.

6. The transaction data processing method according to claim 5, wherein in the decision calculation step, as the decision related to the target, the sum of each column of the P columns of the variation association probability is calculated and sorted to obtain a priority order.

7. The transaction data processing method according to claim 6, wherein in the decision calculation step, as the decision related to the target, n category projects are selected from the P categories and the proportion of the n category projects is calculated respectively.

8. The transaction data processing method according to any one of claims 2 to 7, wherein in the model sample construction step, a consumption proportion vector of a consumer in different industry categories is obtained as the target proportion vector and a model sample is constructed according to the following formula (1), wherein it is assumed that there are P different industries: in the association model construction step, an association model is constructed based on a Markov process according to the following formula (2): Y it = (y it,1 ,y it,2 ,...,y it,P ) T (1) where Y it is the consumption share vector, i denotes the ith sample, t denotes the tth time, and t-1 denotes the t-1th time. in the variation model construction step, a transition probability variation model is constructed according to the following formula (3): Y it = WY it-1 + ε it (2) where W = (w p1p2 ) ∈ R P×P is the correlation probability, w p1p2 represents the transfer probability of consumer's consumption tendency in industry p2 to industry p1, and ε it is a noise term. ​ Y it = WY it-1 + Treat i * ΩY it-1 + ε it (3) Wherein, Ω = (τ p1p2 ) ∈ R P×P is the probability variation matrix, τ p1p2 represents the transition probability of the consumer's consumption variation in industry p2 to industry p1, ε it is a noise term, Treat i represents the action object identifier. 9.The transaction data processing method of claim 8, wherein, In the association model construction step, W is trained such that the error of WY it-1 +ε it is minimized with respect to Y it . 10.The transaction data processing method of claim 9, wherein, in the decision calculating step, column-wise summation is performed on the probability variation matrix and the result is sorted to obtain a priority ranking of the industries for the coupon distribution. 11.The transaction data processing method of claim 8, wherein, in the decision calculating step, n industries are selected to calculate the proportion of the n industries and the proportion is used as the proportion of the industries for the coupon distribution.

12. A transaction data processing system characterized by comprising: comprise: a model sample constructing module configured to obtain a target proportion vector of a target in different item categories and construct a target proportion vector model sample based on the target proportion vector; an association model constructing module configured to construct an association model based on the target proportion vector model sample and an association probability of the target from one item category to another item category; a variation model constructing module configured to construct a variation model based on the target proportion vector model sample, the association probability and a variation association probability of the target from one item category to another item category; a parameter fitting module configured to perform parameter fitting on the variation model based on the target proportion vector model sample to obtain the variation association probability; and a decision calculating module configured to calculate a decision related to the target for different item categories based on the variation association probability. 13.The transaction data processing system of claim 12, wherein, in the variation model constructing module, the variation model is constructed based on the target proportion vector model sample, the association probability, the variation association probability and an action object identifier used to identify whether to participate in a transaction activity, in the variation model constructing step, action samples participating in the transaction activity and non-action samples not participating in the transaction activity are included in a Markov process to construct the variation model. 14.The transaction data processing system of claim 13, wherein, in the association model constructing module, the association model is constructed based on a Markov process. 15.The transaction data processing system of claim 14, wherein, in the parameter fitting module, parameter fitting is performed using any one of the following fitting models: a least square estimation fitting model, a maximum likelihood estimation fitting model, a moment estimation fitting model, a least absolute deviation estimation fitting model and a generalized moment estimation fitting model. 16.The transaction data processing system of claim 15, wherein, when there are P different item categories, the association probability is represented by a P×P dimensional association matrix and the variation association probability is represented by a P×P dimensional association matrix, where P is a natural number. 17.The transaction data processing system of claim 16, wherein, in the decision calculating module, as the decision related to the target, the sum of each column of the P columns of the variation association probability is calculated and sorted to obtain a priority ranking. 18.The transaction data processing system of claim 17, wherein, ​ In the decision calculation module, as the decision related to the target, n category items in P categories are selected, and the proportion of the n category items is calculated respectively.

19. The transaction data processing system according to any one of claims 13 to 18, wherein In the model sample construction module, a consumption proportion vector of the consumer in different industry categories is obtained as the target proportion vector, and a model sample is constructed according to the following formula (1), wherein it is assumed that there are P different industries: Y it = (y it,1 ,y it,2 ,...,y it,P ) T (1) where Y it is the consumption share vector, i denotes the ith sample, t denotes the tth time, and t-1 denotes the t-1th time. In the correlation model construction module, a correlation model is constructed according to the following formula (2) based on a Markov process: Y it = WY it-1 + ε it (2) where W = (w p1p2 ) ∈ R P×P is the correlation probability, w p1p2 denotes the shift probability of the consumer's consumption inclination from industry p2 to industry p1, and εit is a noise term. In the variation model construction module, a transition probability variation model is constructed according to the following formula (3): Y it = WY it-1 + Treat i * ΩY it-1 + ε it (3) wherein Ω = (τ p1p2 ) ∈ R P×P is the probability variation matrix, τ p1p2 represents the transition probability of the consumer's consumption variation in industry p2 to industry p1, ε it is a noise term, Treat i represents the action object identifier.

20. The transaction data processing system according to claim 19, wherein In the correlation model construction module, W is trained such that the error of WY it-1 +ε it is minimum with respect to Y it .

21. The transaction data processing system according to claim 20, wherein In the decision calculation module, column addition is performed on the probability variation matrix, and a priority ranking is obtained as a ranking of the industry for the coupon distribution.

22. The transaction data processing system according to claim 21, wherein In the decision calculation module, n industries are selected to calculate the proportion of the n industries and the proportion of the industries for the coupon distribution.

23. A computer readable medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the transaction data processing method according to any one of claims 1 to 11.

24. A computer device, comprising a storage module, a processor, and a computer program stored in the storage module and executable on the processor, wherein The processor executes the computer program to implement the transaction data processing method according to any one of claims 1 to 11.

25. A computer program product, comprising a computer program, wherein The computer program is executed by the processor to implement the transaction data processing method according to any one of claims 1 to 11.

Images