TWM568442U - Cash flow grouping system - Google Patents

Abstract

A gold flow grouping system comprises a storage module and a processing module, wherein the storage module stores a plurality of enterprise financial information related to the status of the gold flow of the plurality of enterprise groups, and each enterprise gold flow information includes at least one Corresponding to the enterprise data of at least one pre-existing enterprise, each enterprise data includes multiple remittance information, each remittance information has a corresponding remittance enterprise remitted or remitted by the pre-existing enterprise and a remittance amount; the processing module is electrically connected The storage module, for each enterprise gold flow information, the processing module obtains a total amount of gold flow and a total number of remittances according to the enterprise gold flow information, and according to each enterprise gold flow information, as a plurality of group attributes The total amount of the gold flow and the total number of remittances divide the enterprise financial flow information into multiple clusters.

Description

Golden flow grouping system

The present invention relates to a system for automatically classifying multiple gold flow information into multiple clusters, and in particular to a golden flow grouping system.

When facing the credit business of corporate customers, the financial industry must immediately understand the customer's financial information. However, the existing information analysis of the golden stream is based on manual methods to define multiple classification conditions, for example, within one year. Remittance amount, number of times, etc., and then divide the huge customer data into multiple clusters according to the classification conditions, but the labor and time cost of the manual classification method is too much, which is also a big burden for the banking industry.

In view of this, how to provide an artificial intelligence method to automatically analyze the classification conditions, classify the customer data into a plurality of clusters each having the same classification condition, and then formulate a corresponding strategy for the customer group having the same classification condition, that is, The primary topic of creation is to solve.

Therefore, the purpose of the present invention is to provide an automatic multi-flow of gold Information is classified into clustered systems of multiple clusters.

Therefore, the novel gold flow grouping system comprises a storage module and a processing module electrically connected to the storage module.

The storage module stores a plurality of enterprise financial information related to the status of the golden flow of the plurality of enterprise groups, and each enterprise financial flow information includes at least one enterprise data corresponding to at least one pre-stored enterprise, and each enterprise information includes more Remittance information, each remittance information has a corresponding remittance enterprise remitted or remitted by the pre-existing enterprise and a remittance amount.

For each enterprise financial flow information, the processing module obtains a total amount of gold flow and a total number of remittances according to the enterprise financial flow information, and the total amount of the golden flow corresponding to each grouping attribute corresponding to each enterprise gold flow information. And the total number of remittances, the company's financial flow information is divided into multiple clusters.

The effect of the novel is that the processing module automatically classifies the enterprise financial information into the clusters by using the total amount of the golden flows obtained by the enterprise financial information and the total number of the remittances, so that the analysis Personnel can plan a corresponding strategy for these clusters.

100‧‧‧Goldstream Cluster System

1‧‧‧ storage module

2‧‧‧Processing module

51~56‧‧‧Steps

561’~563’‧‧‧ substeps

61~67‧‧‧Steps

621~622‧‧‧ substeps

641~642‧‧‧Substeps

Other features and effects of the present invention will be apparent from the following description of the drawings, in which: 1 is a block diagram showing an embodiment of the implementation of the present novel golden stream grouping system; FIG. 2 is a flow chart illustrating the execution of a clustering attribute obtaining program in the gold flow grouping method of the embodiment; FIG. FIG. 4 is a schematic diagram showing an undirected graph in which one of the golden stream grouping methods is converted by the gold flow grouping method in the embodiment; FIG. Is a schematic diagram illustrating another golden flow relationship diagram of the method for performing the golden flow grouping method; FIG. 6 is a flowchart illustrating how the cluster execution attribute obtaining program obtains another golden stream transfer degree; FIG. Is a flow chart illustrating that the embodiment performs a clustering cluster obtaining procedure in the golden stream grouping method; FIG. 8 is a flowchart illustrating how the clustering obtaining the non-initial center enterprise information is performed by the embodiment. The detailed flow; and FIG. 9 is a flow chart illustrating the detailed flow of how the clustering acquisition program of the embodiment classifies each enterprise information.

Referring to FIG. 1, an embodiment of the novel gold flow grouping system 100 includes a The storage module 1 and a processing module 2 electrically connected to the storage module 1.

The storage module 1 stores a plurality of enterprise financial information related to the status of the golden flow of the plurality of enterprise groups, and each enterprise financial flow information includes at least one enterprise data corresponding to at least one pre-stored enterprise, and each enterprise information includes A plurality of remittance information, each remittance information has a corresponding remittance enterprise remitted or remitted by the pre-existing enterprise, a remittance amount, and a remittance time point corresponding to the remittance amount.

Each enterprise gold flow information stored in the storage module 1 can be represented as a golden flow relationship diagram, and each golden flow relationship diagram includes a plurality of nodes and a plurality of directed edges related to the links between the nodes Each node corresponds to at least one pre-stored enterprise and one of the at least one remittance enterprise, and the at least one pre-existing enterprise and the at least one remittance remitted or remitted by at least one of the directed edges At least one of the other nodes of the enterprise is connected, wherein each directed edge indicates that the connection between the two nodes is a remittance or remittance.

In this embodiment, the embodiment of the golden current grouping system 100 is, for example, a personal computer and a server, but is not limited thereto.

The following is a description of the operation of the storage module 1 of the golden flow grouping system 100 and the operation details of the components of the processing module 2 by performing a golden flow grouping method in the embodiment of the present golden flow grouping system 100. The clustering method includes a clustering attribute acquisition program and a clustering cluster acquisition procedure.

Referring to FIG. 2, the group attribute obtaining program is used to obtain a plurality of different grouping attributes, and the group attribute obtaining program includes a step 51, a step 52, and a step. 53. A step 54, a step 55, and a step 56 (or a step 56').

In step 51, for each enterprise financial information, the processing module 2 calculates the sum of the remittance amounts of all the remittance information in the enterprise financial information, obtains a pair of enterprise financial information, and serves as a plurality of sub-group attributes. One of the total amount of gold flow.

In step 52, for each enterprise financial flow information, the processing module 2 calculates the quantity of all the remittance information in the enterprise golden flow information, obtains a pair of corporate financial information, and serves as one of the sub-group attributes. The total number of remittances.

In step 53, for each enterprise financial information, the processing module 2 divides the total amount of the golden stream by the total number of remittances, and obtains a pair of corporate financial information, and is one of the attributes of the group. Average remittance amount.

In step 54, for each enterprise financial information, the processing module 2 uses the Ljung-Box Q test algorithm to obtain a pair of corporate financial information according to the remittance information, and indicates the randomness of the enterprise. And as a self-correlation check value of one of the grouping attributes.

In step 55, for each gold flow relationship map corresponding to the enterprise gold flow information, the processing module 2 uses the following summer nutrient entropy calculation formula (the following formula (1)) to obtain a pair of enterprise financial information (for It should be a golden relationship diagram, and as a concentration of one of these sub-group attributes.

In the Xia Nong entropy calculation formula (formula (1) above), H(X) is an entropy value indicating the degree of concentration, a is one of the nodes in the gold flow relationship diagram, and b is the gold flow relationship diagram. The other node in the middle, C is the number of all undirected edges in the _unbalanced graph of the golden flow relationship converted from the golden flow diagram, and X _ab is the sum of all the remittance amounts of the two nodes a and b.

Referring to FIG. 3 and FIG. 4, for example, if the gold flow relationship diagram corresponding to an enterprise gold flow information is the golden flow relationship diagram shown in FIG. 3, FIG. 4 is converted by the golden flow relationship diagram shown in FIG. In the gold flow relationship undirected graph, the number C of all undirected edges in the gold flow relationship undirected graph shown in Fig. 4 is 3.

In step 56, for each gold flow relationship map corresponding to each enterprise gold flow information, the processing module 2 divides the sum of the path lengths of all the different paths in the golden flow relationship diagram by the golden flow. The number of all the paths of the relationship map to obtain a pair of business flow information (corresponding to the gold flow relationship diagram), and the degree of gold flow transmission as one of the group attributes. Wherein, all paths of the golden flow relationship diagram do not include any loop (Cycle).

Referring to FIG. 5, for example, assuming a golden flow relationship diagram corresponding to an enterprise golden flow information (as shown in FIG. 5), all available paths are respectively, A->B, A->B->C, A. ->B->D, B->D, B->C, B->C->A, C->A, C->A->B, and C->A->B->D, A total of 9 paths, and A->B->C->A, B->C->A->B, B->C->A->B->D, and C->A->B ->C is a loop, so it does not belong to the gold flow off All paths to the system. The path length of the path A->B is 1, and the path length of the path C->A->B->D is 3. Therefore, the golden flow relationship of the golden flow relationship diagram (as shown in FIG. 5) is ( 1+2+2+1+1+2+1+2+3)/9=1.67.

It should be particularly noted that in the present embodiment, the degree of gold flow transmission is obtained by performing step 56, but in other embodiments, the gold obtained in step 56 may be obtained via a step 56'. Another degree of golden flow transfer of the degree of flow transfer, and the degree of transfer of the other gold flow obtained in step 56', replaces the degree of transfer of the golden flow obtained in step 56 as one of the attributes of the group.

In step 56', for each enterprise gold flow information, the processing module 2 generates another gold flow relationship map according to all the remittance time points of the enterprise golden flow information, and obtains a corresponding golden flow relationship diagram. The other gold flow is transmitted.

Referring to Figure 6, it is particularly noted that step 56' further includes a sub-step 561', a sub-step 562', and a sub-step 563'.

In sub-step 561', for each enterprise gold flow information, the processing module 2 removes the remittance information corresponding to the remittance time point not within a time interval.

In sub-step 562 ′, for each enterprise flow information processed after step 561 ′, the processing module 2 represents the enterprise golden stream information processed by step 561 ′ as another golden flow relationship diagram. The other golden flow relationship diagram includes a plurality of nodes and a plurality of directed edges related to the links between the nodes, each node corresponding to at least one pre-stored enterprise and one of the at least one remittance enterprise, and the same At least one of the directed sides and the remitted Or a node connection corresponding to the at least one pre-stored enterprise and the at least one other at least one remittance enterprise, wherein each directed edge indicates that the connection between the two nodes is a remittance or remittance.

In sub-step 563', for each of the other gold flow relationship diagrams, the processing module 2 divides the sum of the path lengths of all the different paths in the other gold flow relationship diagram by the other gold flow relationship diagram. The number of all the paths is obtained, and the other golden stream transmission degree corresponding to the enterprise gold flow information (corresponding to another gold flow relationship diagram) processed in step 561' is obtained. Wherein, all paths of the other gold flow relationship diagram do not include any loop.

Referring to FIG. 7, the clustering cluster obtaining program is used to obtain a plurality of different clustering clusters. The clustering cluster obtaining program includes a step 61, a step 62, a step 63, a step 64, a step 65, and a step 66. And a step 67.

In step 61, the processing module 2 randomly selects, from the enterprise financial information, a plurality of enterprise financial information respectively as the initial centers of the clusters.

In step 62, for each of the enterprise financial information, the non-initial central enterprise financial information that is not the initial center, the processing module 2 according to the non-initial central enterprise financial information corresponding to the group attribute and each The non-initial central enterprise financial flow information is classified into one of the clusters by the initial grouping corresponding to the initial group.

Referring to FIG. 8, it is particularly noted that in this embodiment, step 62 further includes a sub-step 621, and a sub-step 622.

In the sub-step 621, for each non-initial central enterprise financial information, the processing module 2 is configured according to the grouping attributes corresponding to the non-initial central enterprise gold stream information and the grouping attributes corresponding to each initial center. Calculate a similarity between the non-initial central enterprise financial flow information and each initial center.

It should be particularly noted that in this embodiment, the similarity in sub-step 621 can be obtained by using the Euclidean distance (formula (2) below), but not limited thereto.

Where Xi is one of the group attributes corresponding to the non-initial center enterprise gold flow information, and Yi is one of the initial centers corresponding to the group attribute and the non-initial center enterprise information flow information center The grouping attribute of the same attribute of the corresponding group attribute, i≧1, d(X, Y) is the similarity between the non-initial center enterprise gold stream information and the one of the initial centers.

In sub-step 622, for each non-initial central enterprise financial information, the processing module 2 obtains a comparison from the initial centers according to the similarity between the non-initial central enterprise financial information and each initial center. The non-initial central enterprise financial flow information corresponds to the target initial center with the highest similarity, and the non-initial central enterprise financial flow information is classified into the cluster corresponding to the target initial center.

In step 63, for each cluster corresponding to the initial center, the process The module 2 re-acquires the center of the cluster and its corresponding plurality of reference cluster attributes according to the group attributes corresponding to each enterprise gold stream information (including all initial centers) classified into the cluster. It should be particularly noted that in this embodiment, for each reference group attribute of the cluster, the reference group attribute may be an average of all group attributes of the same attribute of the cluster in the cluster, but not Limited.

In step 64, for each enterprise gold flow information, the processing module 2 according to the grouping attribute corresponding to the enterprise gold stream information and the benchmark group attributes corresponding to each center, the enterprise gold flow information Classified to one of these clusters.

Referring to FIG. 9, it is particularly noted that in this embodiment, step 64 further includes a sub-step 641 and a sub-step 642.

In sub-step 641, for each enterprise financial information, the processing module 2 calculates the enterprise financial flow according to the grouping attributes corresponding to the enterprise golden stream information and the reference grouping attributes corresponding to each center. A similarity between information and each center.

It should be particularly noted that, in this embodiment, the similarity obtained in sub-step 641 is obtained in the same manner as the similarity in sub-step 621 (that is, can be obtained by the above formula (2)), but Not limited to this.

Where Xi is one of the grouping attributes corresponding to the enterprise information, and Yi is one of the benchmark group attributes corresponding to one of the centers corresponding to the enterprise information The base of the same attribute of the group attribute The group attribute, i≧1, d(X, Y) is the similarity between the enterprise financial information and the one of the centers.

In sub-step 642, for each enterprise financial information, the processing module 2 obtains the highest corresponding to the enterprise golden flow information from the centers according to the similarity between the enterprise golden flow information and each center. The target center of similarity, and classify the enterprise financial flow information into the cluster corresponding to the target center.

In step 65, for each cluster corresponding to the center, the processing module 2 determines whether each enterprise gold flow information classified into the cluster corresponding to the center is associated with each cluster corresponding to the initial center. The information of a company's gold flow is exactly the same. When the processing module 2 determines that each enterprise gold stream information classified into the cluster corresponding to the center is not exactly the same as each enterprise gold stream information classified into the cluster corresponding to the initial center, the process step 66 is performed. When the processing module 2 determines that each enterprise gold stream information classified into the cluster corresponding to the center is identical to each enterprise gold stream information classified into the cluster corresponding to the initial center, proceed to process step 67. .

In step 66, the processing module 2 takes the center as the initial center, and performs step 63, step 64, and step 65 again.

In step 67, the processing module 2 obtains a cluster corresponding to each center.

It should be particularly noted that, in this embodiment, the clustering cluster obtaining program only needs the total amount of the gold stream obtained in step 51, and the total number of remittances obtained in step 52 can be completely executed as the group attribute. No need to go from step 53 to step 56 other group attributes obtained.

It is worth mentioning that, in implementation, after obtaining the cluster corresponding to each center, the analysts can find out the potential credit customers, potential supply chain center factory customers according to the common characteristics of each cluster, or Master the status of the corporate stream and the status of repayments. In addition, since the enterprise gold flow information classified into the same cluster has similar clustering attributes, the industrial categories corresponding to each enterprise gold flow information in the same cluster will be the same, and by analyzing each enterprise financial flow information, The cash flow status of the number of remittances and the amount of remittances can be classified into the industry categories corresponding to the enterprise's financial information. For example, the members of the cluster corresponding to the flow of gold flows, but each time the amount of money is small, should be tourism tourism. In this way, the industry categories to which each cluster belongs can be summarized to plan different corresponding strategies for each cluster.

In summary, the novel golden flow grouping system obtains the total amount of the gold flow belonging to the group attribute, the total number of remittances, the average remittance amount, the self-correlation verification value, and the concentration by executing the group attribute obtaining program. Degree, and the degree of the golden flow transmission or the degree of the other golden flow transmission, and then, by means of artificial intelligence, the program is obtained through the clustering cluster, and according to the clustering attributes, the cluster corresponding to each center is obtained, and the analyst can The cluster plans a corresponding strategy. Therefore, it is indeed possible to achieve the purpose of the present invention.

However, the above is only an embodiment of the present invention, and the scope of the present invention cannot be limited thereto, and the patent scope and patent specification are applicable according to the present invention. The simple equivalent changes and modifications made by the content are still within the scope of this new patent.

Claims (9)

A golden flow grouping system, comprising: a storage module, storing a plurality of enterprise golden flow information respectively related to the status of the golden flow of the plurality of enterprise groups, each enterprise golden flow information comprising at least one corresponding to at least one pre-stored enterprise respectively Enterprise information, each enterprise information includes multiple remittance information, each remittance information has a corresponding remittance enterprise remitted or remitted by the pre-existing enterprise and a remittance amount; and a processing module electrically connected to the storage module, For each enterprise financial flow information, the processing module obtains a total amount of gold flow and a total number of remittances according to the enterprise financial flow information, and according to the total amount of the golden stream corresponding to each group attribute of each enterprise gold flow information and The total number of remittances divides the enterprise financial flow information into multiple clusters. The gold flow grouping system of claim 1, wherein the processing module randomly selects, from the enterprise financial information, a plurality of enterprise financial information respectively as initial centers of the clusters, for the enterprises Each of the financial information is not used as the initial center of the non-initial central enterprise financial information, and the processing module according to the non-initial central enterprise financial flow information corresponding to the group attribute and the corresponding group attribute corresponding to each initial center And classifying the non-initial central enterprise financial flow information into one of the clusters, and for each cluster corresponding to the initial center, the processing module is configured according to each enterprise gold flow information classified into the cluster Equally grouping attributes, regaining the center of the cluster and its corresponding plurality of benchmark group attributes, for each enterprise stream information, the processing module according to the enterprise gold stream information corresponding to the group attribute and each center Corresponding to these benchmark group attributes, the company The Golden Stream information is classified into one of the clusters, and for each cluster corresponding to the center, the processing module determines whether each enterprise gold flow information classified into the cluster corresponding to the center is classified to the initial Each enterprise gold flow information of the cluster corresponding to the center is completely the same, when the processing module determines each enterprise gold flow information classified into the cluster corresponding to the center and each cluster corresponding to the cluster corresponding to the initial center When the enterprise financial flow information is not completely the same, the center is taken as the initial center, and the center of the cluster corresponding to each initial center and its corresponding multiple reference group attributes are obtained according to each initial center, and each The enterprise financial flow information is classified into one of the clusters, and it is re-determined whether each enterprise financial flow information of the cluster corresponding to each initial center is identical to each enterprise golden flow information of the cluster corresponding to each center. When the processing module determines each enterprise gold flow information classified into a cluster corresponding to the center and each cluster corresponding to the cluster corresponding to the initial center When the industry cash flow information identical to each center to obtain the corresponding cluster. The golden flow grouping system of claim 2, wherein, for each non-initial central enterprise financial flow information, the processing module is based on the grouping attributes corresponding to the non-initial central enterprise gold flow information and each initial center Corresponding to the group attribute, calculating a similarity between the non-initial center enterprise gold flow information and each initial center, and for each non-initial center enterprise gold flow information, the processing module is based on the non-initial center enterprise gold flow information and The similarity between each initial center, from the initial centers, obtains a target initial center corresponding to the non-initial central enterprise gold flow information, and classifies the non-initial central enterprise financial flow information to the initial target in In the cluster corresponding to the heart, for each enterprise gold flow information, the processing module calculates the enterprise golden flow according to the grouping attributes corresponding to the enterprise gold stream information and the benchmark group attributes corresponding to each center. A similarity between the information and each center. For each enterprise information, the processing module obtains a golden flow from the center according to the similarity between the enterprise information and each center. The information corresponds to the target center with the highest similarity, and the enterprise financial flow information is classified into the cluster corresponding to the target center. The golden flow grouping system according to claim 1, wherein, for each enterprise gold flow information, the total amount of the gold flow corresponding to the enterprise golden flow information is calculated by the processing module to calculate all the remittances in the enterprise golden flow information. The sum of the remittance amount of the information is obtained, and the total remittance number corresponding to the enterprise financial flow information is obtained by the processing module calculating the amount of all remittance information in the enterprise golden flow information. The golden flow grouping system according to claim 1, wherein each enterprise gold flow information is represented as a golden flow relationship diagram, and each golden flow relationship diagram includes a plurality of nodes and a plurality of links related to the nodes a directed edge, each node corresponding to at least one pre-existing enterprise and one of the at least one remittance enterprise, and at least one of the directed parties and the at least one pre-existing enterprise remitted or remitted At least one of the at least one remittance enterprise corresponds to a node connection, wherein each directed edge indicates that the connection between the two nodes is a remittance or remittance. For each golden flow relationship diagram, the processing module utilizes the following summer The agricultural entropy calculation formula obtains a degree of concentration, and the concentration degree is used as another sub-group attribute of the enterprise gold flow information corresponding to the golden flow relationship map: Where H(X) is an entropy value indicating the degree of concentration, a is one of the nodes in the gold flow diagram, b is another node in the gold flow diagram, and C is a gold stream The number of all undirected edges in the undirected graph of the golden flow relationship converted by the relation graph, and X _ab is the sum of all the remittance amounts of the two nodes a and b, and the processing module not only according to the total amount of the golden stream and the total remittance The number of times, according to the degree of concentration, the enterprise financial flow information is divided into multiple clusters. The golden stream grouping system of claim 1, wherein each enterprise gold stream information is represented as a corresponding gold flow relationship diagram, the gold flow relationship diagram comprising a plurality of nodes and a plurality of related nodes a directed edge of the link, each node corresponding to at least one pre-stored enterprise and one of the at least one remittance enterprise, and the at least one pre-stored enterprise remitted or remitted by at least one of the directed edges And a node connection corresponding to at least one of the at least one money transfer enterprise, wherein each directed edge indicates that the connection between the two nodes is a remittance or import, and for each gold flow relationship diagram, the processing module The sum of the path lengths of all the different paths in the golden flow diagram, divided by the number of all the paths of the golden flow diagram to obtain a golden flow transfer degree, wherein all paths of the golden flow relationship diagram do not contain any loop, And the golden stream transfer degree is used as another group attribute of the enterprise gold flow information corresponding to the gold flow relationship diagram, and the processing module not only according to the total amount of the golden stream and the total number of remittances, but also according to the golden stream transfer degree , such Enterprise Golden Stream information is divided into multiple clusters. The gold flow grouping system of claim 1, wherein each remittance information further has a remittance time point corresponding to the remittance amount, and for each enterprise gold flow information, the processing module will correspond to a remittance time point. The remittance information is removed in a time interval. For each enterprise financial flow information, the processing module represents the enterprise golden flow information as a golden flow relationship diagram, the golden flow relationship diagram includes a plurality of nodes and a plurality of related a directed edge of the connection between the nodes, each node corresponding to at least one pre-stored enterprise and one of the at least one remittance enterprise, and at least one of the directed edges and the remitted or remitted a node corresponding to at least one of the pre-stored enterprise and the at least one of the at least one remittance enterprise, wherein each directed edge indicates that the connection between the two nodes is a remittance or import, and for each golden flow relationship diagram, The processing module divides the sum of the path lengths of all the different paths in the golden flow diagram by the number of all the paths of the golden flow graph to obtain another golden flow transfer degree, wherein all paths of the golden flow relationship map Does not include any loop, and the other golden stream transfer degree is another group attribute of the enterprise gold stream information corresponding to the gold flow relationship graph, and the processing module not only according to the total amount of the gold stream and the total number of remittances And according to the degree of the other golden stream transmission, the enterprise gold flow information is divided into a plurality of clusters. The gold flow grouping system of claim 1, wherein each remittance information further has a remittance time point corresponding to the remittance amount, and for each enterprise gold flow information, the processing module utilizes Ljung according to the remittance information. The -Box Q test algorithm obtains a self-correlation verification value indicating the randomness of the enterprise, and the processing module not only uses the total amount of the golden stream and the total number of remittances, but also the enterprise data according to the self-related verification value. Divided into these groups set. The golden stream grouping system of claim 1, wherein, for each enterprise gold stream information, the processing module divides the total amount of the gold stream by the total number of remittances to obtain an average remittance amount, and the processing module is not only based on The total amount of the gold flow and the total number of remittances are also divided into the clusters based on the average remittance amount.