CN114066173A - Capital flow behavior analysis method and storage medium - Google Patents
Capital flow behavior analysis method and storage medium Download PDFInfo
- Publication number
- CN114066173A CN114066173A CN202111247487.XA CN202111247487A CN114066173A CN 114066173 A CN114066173 A CN 114066173A CN 202111247487 A CN202111247487 A CN 202111247487A CN 114066173 A CN114066173 A CN 114066173A
- Authority
- CN
- China
- Prior art keywords
- flow
- fund
- fund flow
- group
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 25
- 238000000034 method Methods 0.000 claims abstract description 34
- 230000006399 behavior Effects 0.000 claims abstract description 29
- 238000013136 deep learning model Methods 0.000 claims abstract description 10
- 238000004590 computer program Methods 0.000 claims description 13
- 230000002159 abnormal effect Effects 0.000 description 12
- 230000008569 process Effects 0.000 description 9
- 238000012544 monitoring process Methods 0.000 description 7
- 238000012550 audit Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 230000005856 abnormality Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000875 corresponding effect Effects 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010224 classification analysis Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000005206 flow analysis Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/103—Workflow collaboration or project management
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Operations Research (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Marketing (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Medical Informatics (AREA)
- Mathematical Optimization (AREA)
- Development Economics (AREA)
- Mathematical Analysis (AREA)
- Educational Administration (AREA)
- Computational Mathematics (AREA)
- Game Theory and Decision Science (AREA)
- Algebra (AREA)
- Probability & Statistics with Applications (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
A capital flow behavior analysis method and a storage medium, wherein the method comprises the following steps: receiving capital flow data, dividing the capital flow data into a first group, and judging the capital flow data of the first group through a first judgment rule, wherein the method comprises the following steps: calculating a capital flow equivalent for the first group, denoted Ai，
Applying a Bayesian probability-based deep learning model based on probability distribution of capital flow equivalentsObtain the threshold AuThe capital flow equivalent Ai is compared with a threshold value AuComparing, if Ai is more than or equal to AuThen it is determined that a fund flow anomaly exists. By the scheme, intelligent analysis of capital flow data can be realized, and analysis efficiency is improved.
Description
Technical Field
The invention belongs to the field of data analysis, and particularly relates to a risk analysis method for enterprise capital flow data.
Background
Enterprises can not avoid various types of fund flow in the operation process, how to quickly and efficiently discover abnormal conditions in the fund flow process, reduce related risks in the enterprise operation process, strengthen the supervision on various fund uses, and adopt periodic manual audit in most enterprises; or the fund flow is examined and approved in multiple levels through a corresponding fund management system, and meanwhile, the monitoring is carried out through establishing rules such as single amount, accumulated amount, flow times and the like, and setting thresholds and the like.
The manual audit workload is large and the efficiency is low: the manual auditing mode is adopted, so that the workload is huge, the opportunity that the best risk is lost and prevention is carried out is basically carried out afterwards, the threshold is set for monitoring, the monitoring is easy to avoid, and the flexibility is lacked: the fund flow condition is monitored in the modes of fund flow amount each time, specific period accumulated amount, flow counting times in a specific period and the like, and the monitoring is avoided after the threshold is found by artificially trying for many times; meanwhile, in the mode, threshold setting rules need to be adjusted according to actual operation conditions of enterprises along with the lapse of time, manual auditing or a threshold method based on a small time window mainly describes static attributes of fund flow, and dynamic continuity of actual fund flow is little involved.
Disclosure of Invention
For this reason, it is desirable to provide a technical solution that enables more intelligent analysis of the capital flow data.
To achieve the above object, the applicant provides a method for analyzing a capital movement behavior, comprising the steps of:
receiving funds movement data, dividing the funds movement data into a first group,
and judging the fund flow data of the first group through a first judgment rule, wherein the method comprises the following steps:
calculating a duration Ti of the money flow data from the money flow data within the first class group;
calculating a capital flow interval Δ Ti ═ Δ Ti1, Δ Ti2, …, Δ Tin-1} from the capital flow data within the first class group;
calculating a difference between a subsequent fund flow amount and a previous fund flow amount according to the fund flow data in the first group: Δ Mi ═ Δ Mi1, Δ Mi2, …, Δ Min-1 };
calculating a capital flow equivalent for the first group, denoted Ai，
Obtaining a threshold A by applying a Bayesian probability-based deep learning model according to the probability distribution of the capital flow equivalenceu，
And comparing the fund flow equivalent Ai with a threshold value Au, and if Ai is more than or equal to Au, determining that the fund flow is abnormal.
Specifically, the method further comprises the step of judging whether the complex fund flow mode is judged, and if the complex fund flow mode is judged, judging the fund flow data clustered in the first class through a first judgment rule.
Specifically, if the result of judging whether the fund flow data of the first group is the complex fund flow mode is negative, the judgment is performed according to a second judgment rule:
and comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not.
In particular, the method also comprises the step of,
and judging the fund flow data of the first group by a second judgment rule:
comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not;
and weighting according to the judgment result of the first judgment rule and/or the judgment result of the second judgment rule to obtain the result of whether to carry out early warning.
Specifically, the money flow data includes attributes: one or more of floating time, floating amount, payer, receiver, type of the country or country, fund usage, and currency,
the dividing the fund flow data into the first class groups is to divide the fund flow data with the same plurality of attribute values into the same class groups.
A funds movement behavior analysis storage medium storing a computer program which when executed performs steps comprising:
receiving funds movement data, dividing the funds movement data into a first group,
and judging the fund flow data of the first group through a first judgment rule, wherein the method comprises the following steps:
calculating a duration Ti of the money flow data from the money flow data within the first class group;
calculating a capital flow interval Δ Ti ═ Δ Ti from the capital flow data within the first class group1，ΔTi2,…,ΔTin-1}；
Calculating a difference between a subsequent fund flow amount and a previous fund flow amount according to the fund flow data in the first group: Δ Mi ═ Δ Mi { (Δ Mi)1，ΔMi2,…,ΔMin-1}；
Calculating a capital flow equivalent for the first group, denoted Ai。
Obtaining a threshold A by applying a Bayesian probability-based deep learning model according to the probability distribution of the capital flow equivalenceu。
The capital flow equivalent Ai is compared with a threshold value AuComparing, if Ai is more than or equal to AuThen it is determined that there is a fund flow differenceOften times.
Specifically, the computer program when executed further includes a step of judging whether the complex fund flow mode is determined, and if yes, judging the fund flow data grouped in the first category according to a first judgment rule.
In particular, the computer program, when executed, performs: judging whether the result of the complex fund flow mode is negative or not for the fund flow data of the first group, and judging according to a second judgment rule:
and comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not.
In particular, the computer program when executed further comprises steps,
and judging the fund flow data of the first group by a second judgment rule:
comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not;
and weighting according to the judgment result of the first judgment rule and/or the judgment result of the second judgment rule to obtain the result of whether to carry out early warning.
Optionally, the funds movement data includes attributes: one or more of floating time, floating amount, payer, receiver, type of the country or country, fund usage, and currency,
the dividing the fund flow data into the first class groups is to divide the fund flow data with the same plurality of attribute values into the same class groups.
Different from the prior art, the technical scheme can perform classification analysis on the capital flow data, then utilizes the capital flow equivalent to perform threshold judgment, and finally obtains the judgment result whether the capital flow behavior is abnormal or not.
Drawings
Fig. 1 is a flow chart of a method for analyzing a fund flow behavior according to an embodiment.
Detailed Description
To explain technical contents, structural features, and objects and effects of the technical solutions in detail, the following detailed description is given with reference to the accompanying drawings in conjunction with the embodiments.
The supervision needs to manage and audit the large amount of fund use condition of the enterprise and the enterprise group also needs to manage and audit the fund use condition of the enterprise. The invention provides a capital flow monitoring method based on a model, which models capital flow behaviors of enterprises for research emphasis according to time sequence characteristics of capital flow and extracts each flow behavior characteristic after classifying the capital flow division flow units of the enterprises, thereby enabling quantitative analysis of capital flow to be possible and facilitating discovery of abnormity in the capital flow process. Therefore, the workload of manual audit can be reduced, more importantly, the abuse of enterprise funds is avoided, the risk in the enterprise operation process is greatly reduced, and the competitiveness of the enterprise is improved.
The fund flow is not only a single action, but also is in the process of enterprise operation, the fund flow mode of the enterprise is directly related to the enterprise operation purpose, and the information of fund flow objects, flow time, flow range and the like represents the behavior system of the fund flow of the enterprise. Also, the money flow behavior is a continuous activity, and the relationship with other money flows in the early period or money flows through other channels needs to be considered, so that the continuous monitoring of the money flows is very important. From a data analysis perspective, this means that potential anomalies can be detected by monitoring a large amount of capital flow sequence data, even data spanning years, which is very difficult to detect through manual post-audit.
To further illustrate the capital movement behavior patterns, the capital movement data needs to be analyzed, and in fact, the capital movement of each enterprise in the operation process for achieving the operation target is composed of a plurality of capital movement data records. The funds movement data includes the following attributes: one or more of floating time, floating amount, payer, receiver, type of the country or country, fund usage and currency. The flow time sequence can deduce the time interval, and the specific value of the payer and the receiver can be the name of the business.
Dividing the fund flow data of a certain specific enterprise to achieve the same operation target into a group according to different dimensions, such as enterprise dimensions, fund use dimensions, time interval dimensions and the like, wherein the groups can also be called fund flow units, and the division of the fund flow data into a first group specifically comprises dividing the fund flow data with a plurality of same attribute values into the same group, and marking the same group as' Ui", such that each flow of funds characterizes the commonality of a series of the units of funds flow, and each unit of funds flow also becomes an instance of a certain pattern of funds flow, with corresponding statistical characteristics, and therefore for the same group. The extraction of the statistical features of the fund flow is a sampling of the fund flow pattern recognition, which is denoted as MiThe capital flow S of the enterprise is thus composed of these capital flow units:
S＝{U1，U2,…,Un}
each of the capital movement units UiThe capital flow data in (1) is noted as:
Fi＝{Fi1，Fi2,…,Finattributes of the funds-flow data include: time of fund flow, amount of the flow, payer of the fund, etc.
Accordingly, the pattern of the enterprise user's fund flow may be characterized by the fund flow pattern, namely:
B＝{M1,M2，…，Mn}
meanwhile, we define the duration of the fund flow unit as T, if there is only one fund flow data in the fund flow duration unit, T is 0, if there are multiple fund flow data, T is the difference between the latest fund flow time and the earliest fund flow time in the fund flow unit.
The behavior analysis mode based on the fund flow can be analyzed through the fund flow record, the fund flow unit and the fund flow mode with time sequence characteristics, and the analysis process covers the time and the flow direction of the fund flow, the interval of the fund flow occurrence time of the same fund flow unit, the payment amount, the fund use, the internal and external types, the currency and other attributes.
Referring to fig. 1, the method for analyzing a capital flow behavior of the present embodiment includes the following steps:
s1 receives the funds-flow data, groups the funds-flow data into a first group,
s2, determining the fund flow data of the first group according to a first determination rule, including the steps of:
calculating a duration Ti of the money flow data from the money flow data within the first class group;
calculating a capital flow interval Δ Ti ═ Δ Ti1, Δ Ti2, …, Δ Tin-1} from the capital flow data within the first class group;
calculating a difference between a subsequent fund flow amount and a previous fund flow amount according to the fund flow data in the first group: Δ Mi ═ Δ Mi1, Δ Mi2, …, Δ Min-1 };
calculating the capital flow equivalent of the capital flow unit, marked Ai。
Obtaining a threshold A by applying a Bayesian probability-based deep learning model according to the probability distribution of the capital flow equivalenceu。
And comparing the fund flow equivalent Ai with a threshold value Au, and if Ai is more than or equal to Au, determining that the fund flow is abnormal.
In this embodiment, the fund flow data may be obtained from other data collection methods, may also be obtained by inputting and importing data by a user, or may also be obtained or acquired from a memory or a network by a computer running a fund flow analysis method through a calling program or the like. Then dividing part or all of the fund flow data into a first group according to the requirement, generally speaking, sorting the fund flow data in the first group according to the time sequence; in some embodiments, the duration Ti is the difference between the point in time of the earliest funds-flow data in the first group and the point in time of the latest funds-flow data in the first group. By applying the Bayesian probability-based deep learning model, the technical effect of assisting in judging whether data abnormality exists can be achieved by establishing the Bayesian probability-based deep learning model and learning by using artificially labeled abnormal data with reference to the prior art. According to the scheme, the mutual relation between time and amount of the fund flow data in a group can be evaluated through the independently designed index fund flow equivalent Ai, and the judgment result of whether the data is abnormal or not is automatically obtained in a machine learning mode. The analysis and discrimination work of the set of the fund flow data is improved well, and meanwhile, the attribute is added to the fund flow data more subsequently, so that the angle and the dimensionality of analysis are expanded; and meanwhile, the advantages of high concurrent writing, low query delay and the like are supported. The analysis algorithm adopts a Bayesian algorithm based on statistical characteristics, so that the problem that a service person is difficult to give an accurate and quantitative abnormal threshold in advance can be avoided. Abnormal fund flow identification is carried out by adopting various strategy modes, and balance between the recall ratio and the precision ratio can be carried out according to actual supervision requirements. And the capital flow data flow is adjusted in real time to adapt to the change of the business rule. Meanwhile, the selection of the first group can be changed as required, so that the scheme supports the analysis of the capital flow data by adopting different dimensionalities and can also combine multiple dimensionalities for analysis.
In some other specific embodiments, the method further includes the step of, when determining whether the complex fund flow mode is determined, if so, determining the fund flow data of the first group according to a first determination rule. In some embodiments, the fund flow unit comprises a plurality of fund flow time sequence data which are continuous and involve more payers, and the duration period is longer. Therefore, steps can be further performed to score the complexity of the first class group, the complexity score is positively correlated with dimensions such as data volume, number of payers, duration period and the like, and no matter the complexity scoring mode is obtained through calculation, the manually labeled class group with the complex fund flow mode and the manually labeled class group with the simple fund flow mode can be used for learning through applying a deep learning model based on the Bayesian probability, so that the technical effect of better distinguishing the complex class group from the simple class group can be achieved. The first judgment rule is judged for the complex fund flow mode group, the simple (such as few dimensionalities) judgment rule is judged for the simple group, complex calculation is not needed for all groups, and the operation complexity and the operation cost of the analysis method can be reduced.
In some other specific embodiments, if the result of determining whether the fund flow data of the first group is the complex fund flow pattern is negative, the determination is performed according to a second determination rule: and comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not. That is, if the result of the complex fund flow mode is negative, it indicates that the first class group is determined as a simple class group, so that the determination of whether the fund flow abnormality exists can be performed according to the single-dimension determination rule. The key value of the second determination rule may be changed as needed, and may be that the payment amount of the fund flow data exceeds a threshold, that the time of the fund flow data is less than a preset threshold, that an attribute value of "whether the fund flow data is manually marked as abnormal" is included, or the like. That is, the running cost of the method can be saved better by performing the abnormality determination of the simple rule on the simple class group.
In some other specific embodiments, the method further includes the step of determining, by a second determination rule, the fund flow data of the first group: comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not; and weighting according to the judgment result of the first judgment rule and/or the judgment result of the second judgment rule to obtain the result of whether to carry out early warning. In this embodiment, no matter how the complexity score of the first group is, at least two rules are determined for the first group, and the key value comparison for the fund flow data may be that the payment amount of the fund flow data exceeds a threshold, that the time of the fund flow data is less than a preset threshold, that an attribute value of "whether the fund flow data is manually marked as abnormal" is used, or the like. By designing a plurality of judgment rules, the fund flow abnormity of one enterprise can be detected by any judgment rule or can be monitored, so that the early warning of the abnormity possibility can be set by a union decision strategy or a weight decision strategy of the plurality of judgment rules. The decision strategy proposed by the invention comprises union decision and weight decision, and the actual situation can set a specific decision strategy according to the fund control requirement of an enterprise. Therefore, in some embodiments, the result of whether to perform the pre-warning is obtained according to the weighted sum of the determination results of the first determination rule and the weighted sum of the determination results of the second determination rule. The scheme can achieve the early warning effect of the fund flow data more flexibly and more accurately.
In other specific embodiments, the money flow data includes attributes: the fund flow data is divided into a first group, specifically, the fund flow data with a plurality of same attribute values is divided into the same group. By setting the attributes of the fund flow data and dividing the fund flow data through the attributes of a plurality of dimensions, the data in the same group can have more similarity, and the fund flow behavior can be better analyzed for specific dimensions.
In other embodiments, a method for analyzing a fund flow behavior is provided, the method comprising the steps of:
step 1: and (4) collecting the fund flow data, namely uniformly receiving/collecting different types of fund flow data of different enterprise fund related systems. Step 2: and carrying out time-sequencing processing on the fund flow data, carrying out pretreatment such as cleaning, data standardization, data reordering and the like on the collected various fund flow data, and meanwhile, persisting the processed data into a fund flow data time sequence library. And step 3: the fund flow unit is divided into enterprises according to enterprises, fund application and enterprises according to flow time intervalsThe capital movement data is divided into movement units. Dividing flow units by enterprises and recording the flow units as UeDividing flow units by capital usage into UuFlow units divided by time interval are marked as Ut. And 4, step 4: fund flow behavior pattern pre-recognition: according to the flow unit division result Ue、Uu、UtPre-judging the fund flow behavior mode, wherein the judging mode comprises the following steps: simple flow patterns, complex flow patterns. And 5: if the simple flow mode is adopted, the judgment is carried out based on rules, wherein the rules comprise whether the single fund flow exceeds a threshold value, whether the fund flow with a certain enterprise within the year exceeds the threshold value and the like, and the rules can be set according to the actual conditions of the enterprise. Step 6: in the case of complex fund flow patterns, time sequence feature extraction is carried out, and fund flow behaviors are judged by matching based on a trained fund flow feature library. And 7: and (4) determining the fund flow behavior, and identifying and determining whether the fund flow of the enterprise has abnormal risk based on a plurality of strategies. The scheme can be used for judging the abnormal risk of the fund flow data.
The present solution also provides a fund flow behavior analysis storage medium, which stores a computer program, wherein the computer program when executed performs the steps of:
receiving funds movement data, dividing the funds movement data into a first group,
and judging the fund flow data of the first group through a first judgment rule, wherein the method comprises the following steps:
calculating a duration Ti of the money flow data from the money flow data within the first class group;
calculating a capital flow interval Δ Ti ═ Δ Ti from the capital flow data within the first class group1，ΔTi2,…,ΔTin-1}；
Calculating a difference between a subsequent fund flow amount and a previous fund flow amount according to the fund flow data in the first group: Δ Mi ═ Δ Mi { (Δ Mi)1，ΔMi2,…,ΔMin-1}；
Calculating a capital flow equivalent for the first group, denoted Ai。
Obtaining a threshold A by applying a Bayesian probability-based deep learning model according to the probability distribution of the capital flow equivalenceu。
The capital flow equivalent Ai is compared with a threshold value AuComparing, if Ai is more than or equal to AuThen it is determined that a fund flow anomaly exists.
Specifically, the computer program when executed further includes a step of judging whether the complex fund flow mode is determined, and if yes, judging the fund flow data grouped in the first category according to a first judgment rule.
In particular, the computer program, when executed, performs: judging whether the result of the complex fund flow mode is negative or not for the fund flow data of the first group, and judging according to a second judgment rule:
and comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not.
In particular, the computer program when executed further comprises steps,
and judging the fund flow data of the first group by a second judgment rule:
comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not;
and weighting according to the judgment result of the first judgment rule and/or the judgment result of the second judgment rule to obtain the result of whether to carry out early warning.
Optionally, the funds movement data includes attributes: one or more of floating time, floating amount, payer, receiver, type of the country or country, fund usage, and currency,
the dividing the fund flow data into the first class groups is to divide the fund flow data with the same plurality of attribute values into the same class groups.
It should be noted that, although the above embodiments have been described herein, the invention is not limited thereto. Therefore, based on the innovative concepts of the present invention, the technical solutions of the present invention can be directly or indirectly applied to other related technical fields by making changes and modifications to the embodiments described herein, or by using equivalent structures or equivalent processes performed in the content of the present specification and the attached drawings, which are included in the scope of the present invention.
Claims (10)
1. A method of fund flow behavior analysis, comprising the steps of:
receiving funds movement data, dividing the funds movement data into a first group,
and judging the fund flow data of the first group through a first judgment rule, wherein the method comprises the following steps:
calculating a duration Ti of the money flow data from the money flow data within the first class group;
calculating a capital flow interval Δ Ti ═ Δ Ti from the capital flow data within the first class group1，ΔTi2,…,ΔTin-1}；
Calculating a difference between a subsequent fund flow amount and a previous fund flow amount according to the fund flow data in the first group: Δ Mi ═ Δ Mi { (Δ Mi)1，ΔMi2,…,ΔMin-1}；
Calculating a capital flow equivalent for the first group, denoted Ai，
Obtaining a threshold A by applying a Bayesian probability-based deep learning model according to the probability distribution of the capital flow equivalenceu，
The capital flow equivalent Ai is compared with a threshold value AuComparing, if Ai is more than or equal to AuThen it is determined that a fund flow anomaly exists.
2. The method of fund flow behavior analysis according to claim 1, further comprising the step of determining whether the complex fund flow pattern is determined, and if so, determining the first class of the fund flow data according to a first determination rule.
3. The method according to claim 2, wherein the result of determining whether the fund flow data of the first group is a complex fund flow pattern is negative, and then the determination is made according to a second determination rule:
and comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not.
4. The money flow behavior analysis method according to claim 1, further comprising a step of,
and judging the fund flow data of the first group by a second judgment rule:
comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not;
and weighting according to the judgment result of the first judgment rule and/or the judgment result of the second judgment rule to obtain the result of whether to carry out early warning.
5. The fund flow behavior analysis method according to claim 1, wherein the fund flow data comprises attributes: one or more of floating time, floating amount, payer, receiver, type of the country or country, fund usage, and currency,
the dividing the fund flow data into the first class groups is to divide the fund flow data with the same plurality of attribute values into the same class groups.
6. A funds movement behavior analysis storage medium, having stored thereon a computer program which, when executed, performs steps comprising:
receiving funds movement data, dividing the funds movement data into a first group,
and judging the fund flow data of the first group through a first judgment rule, wherein the method comprises the following steps:
calculating a duration Ti of the money flow data from the money flow data within the first class group;
calculating a capital flow interval Δ Ti ═ Δ Ti from the capital flow data within the first class group1，ΔTi2,…,ΔTin-1}；
Calculating a difference between a subsequent fund flow amount and a previous fund flow amount according to the fund flow data in the first group: Δ Mi ═ Δ Mi { (Δ Mi)1，ΔMi2,…,ΔMin-1}；
Calculating a capital flow equivalent for the first group, denoted Ai。
Obtaining a threshold A by applying a Bayesian probability-based deep learning model according to the probability distribution of the capital flow equivalenceu。
The capital flow equivalent Ai is compared with a threshold value AuComparing, if Ai is more than or equal to AuThen it is determined that a fund flow anomaly exists.
7. The funds-flow behavior analysis storage medium of claim 6, wherein the computer program when executed performs further comprising the step of determining whether the determination is a complex funds-flow pattern, and if so, determining the first type of funds-flow data based on a first determination rule.
8. The funds-flow behavior analysis storage medium of claim 7, wherein the computer program when executed performs: judging whether the result of the complex fund flow mode is negative or not for the fund flow data of the first group, and judging according to a second judgment rule:
and comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not.
9. The funds-flow behavior analysis storage medium of claim 6, wherein the computer program when executed performs further comprising the steps of,
and judging the fund flow data of the first group by a second judgment rule:
comparing key values of the fund flow data, and judging whether fund flow abnormity exists or not;
and weighting according to the judgment result of the first judgment rule and/or the judgment result of the second judgment rule to obtain the result of whether to carry out early warning.
10. The funds-flow behavior analysis storage medium of claim 6, wherein the funds-flow data comprises attributes: one or more of floating time, floating amount, payer, receiver, type of the country or country, fund usage, and currency,
the dividing the fund flow data into the first class groups is to divide the fund flow data with the same plurality of attribute values into the same class groups.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111247487.XA CN114066173A (en) | 2021-10-26 | 2021-10-26 | Capital flow behavior analysis method and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111247487.XA CN114066173A (en) | 2021-10-26 | 2021-10-26 | Capital flow behavior analysis method and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114066173A true CN114066173A (en) | 2022-02-18 |
Family
ID=80235505
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111247487.XA Pending CN114066173A (en) | 2021-10-26 | 2021-10-26 | Capital flow behavior analysis method and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114066173A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116167852A (en) * | 2023-04-19 | 2023-05-26 | 苏州傲林科技有限公司 | Method and device for processing fund flow direction data |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050222929A1 (en) * | 2004-04-06 | 2005-10-06 | Pricewaterhousecoopers Llp | Systems and methods for investigation of financial reporting information |
| CN104021456A (en) * | 2014-06-23 | 2014-09-03 | 国网山东省电力公司 | Method and system used for all-around enterprise group capital data monitoring |
| US20170157439A1 (en) * | 2014-07-08 | 2017-06-08 | Amrona Ag | Device for providing a crossover point |
| CN108280755A (en) * | 2018-02-28 | 2018-07-13 | 阿里巴巴集团控股有限公司 | The recognition methods of suspicious money laundering clique and identification device |
| CN109801073A (en) * | 2018-12-13 | 2019-05-24 | 中国平安财产保险股份有限公司 | Risk subscribers recognition methods, device, computer equipment and storage medium |
| CN110493180A (en) * | 2019-07-05 | 2019-11-22 | 湖北工业大学 | A kind of substation network communication flow real-time analysis method |
| CN112017061A (en) * | 2020-07-15 | 2020-12-01 | 北京淇瑀信息科技有限公司 | Financial risk prediction method and device based on Bayesian deep learning and electronic equipment |
| CN112801456A (en) * | 2020-12-30 | 2021-05-14 | 北京知因智慧科技有限公司 | Fund liquidity risk early warning method, device and electronic equipment |
| CN112801800A (en) * | 2021-04-14 | 2021-05-14 | 深圳格隆汇信息科技有限公司 | Behavior fund analysis system, behavior fund analysis method, computer equipment and storage medium |
| CN112905671A (en) * | 2021-03-24 | 2021-06-04 | 北京必示科技有限公司 | Time series exception handling method and device, electronic equipment and storage medium |
| CN113450218A (en) * | 2021-06-30 | 2021-09-28 | 福建华通银行股份有限公司 | Client fund state tracking design method based on time axis change |
-
2021
- 2021-10-26 CN CN202111247487.XA patent/CN114066173A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050222929A1 (en) * | 2004-04-06 | 2005-10-06 | Pricewaterhousecoopers Llp | Systems and methods for investigation of financial reporting information |
| CN104021456A (en) * | 2014-06-23 | 2014-09-03 | 国网山东省电力公司 | Method and system used for all-around enterprise group capital data monitoring |
| US20170157439A1 (en) * | 2014-07-08 | 2017-06-08 | Amrona Ag | Device for providing a crossover point |
| CN108280755A (en) * | 2018-02-28 | 2018-07-13 | 阿里巴巴集团控股有限公司 | The recognition methods of suspicious money laundering clique and identification device |
| CN109801073A (en) * | 2018-12-13 | 2019-05-24 | 中国平安财产保险股份有限公司 | Risk subscribers recognition methods, device, computer equipment and storage medium |
| CN110493180A (en) * | 2019-07-05 | 2019-11-22 | 湖北工业大学 | A kind of substation network communication flow real-time analysis method |
| CN112017061A (en) * | 2020-07-15 | 2020-12-01 | 北京淇瑀信息科技有限公司 | Financial risk prediction method and device based on Bayesian deep learning and electronic equipment |
| CN112801456A (en) * | 2020-12-30 | 2021-05-14 | 北京知因智慧科技有限公司 | Fund liquidity risk early warning method, device and electronic equipment |
| CN112905671A (en) * | 2021-03-24 | 2021-06-04 | 北京必示科技有限公司 | Time series exception handling method and device, electronic equipment and storage medium |
| CN112801800A (en) * | 2021-04-14 | 2021-05-14 | 深圳格隆汇信息科技有限公司 | Behavior fund analysis system, behavior fund analysis method, computer equipment and storage medium |
| CN113450218A (en) * | 2021-06-30 | 2021-09-28 | 福建华通银行股份有限公司 | Client fund state tracking design method based on time axis change |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116167852A (en) * | 2023-04-19 | 2023-05-26 | 苏州傲林科技有限公司 | Method and device for processing fund flow direction data |
| CN116167852B (en) * | 2023-04-19 | 2023-07-04 | 苏州傲林科技有限公司 | Method and device for processing fund flow direction data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111882446B (en) | Abnormal account detection method based on graph convolution network | |
| CN107633030B (en) | Credit evaluation method and device based on data model | |
| CN109886284B (en) | Fraud detection method and system based on hierarchical clustering | |
| CN111178675A (en) | LR-Bagging algorithm-based electric charge recycling risk prediction method, system, storage medium and computer equipment | |
| US20200286095A1 (en) | Method, apparatus and computer programs for generating a machine-learning system and for classifying a transaction as either fraudulent or genuine | |
| CA3120412A1 (en) | An automated and dynamic method and system for clustering data records | |
| CN114707571B (en) | Credit data anomaly detection method based on enhanced isolation forest | |
| CN110930038A (en) | Loan demand identification method, loan demand identification device, loan demand identification terminal and loan demand identification storage medium | |
| CN107633455A (en) | Credit estimation method and device based on data model | |
| CN110083507A (en) | Key Performance Indicator classification method and device | |
| Ruyu et al. | A comparison of credit rating classification models based on spark-evidence from lending-club | |
| AbdulSattar et al. | Fraudulent transaction detection in FinTech using machine learning algorithms | |
| CN115357764A (en) | Abnormal data detection method and device | |
| CN114066173A (en) | Capital flow behavior analysis method and storage medium | |
| CN113506160A (en) | Risk early warning method and system for unbalanced financial text data | |
| CN112329862A (en) | Decision tree-based anti-money laundering method and system | |
| CN111046947B (en) | Training system and method of classifier and recognition method of abnormal sample | |
| CN112465397A (en) | Audit data analysis method and device | |
| CN112348685A (en) | Credit scoring method, device, equipment and storage medium | |
| Himani et al. | A comparative study on machine learning based prediction of citations of articles | |
| Budianto et al. | Machine learning-based approach on dealing with binary classification problem in imbalanced financial data | |
| Yu et al. | An automatic recognition method of journal impact factor manipulation | |
| Nikolikj et al. | Sensitivity Analysis of RF+ clust for Leave-one-problem-out Performance Prediction | |
| US20200285895A1 (en) | Method, apparatus and computer program for selecting a subset of training transactions from a plurality of training transactions | |
| CN114626940A (en) | Data analysis method and device and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |