CN110991695A - Regional high-risk confidence losing behavior prediction method based on enterprise relation map - Google Patents
Regional high-risk confidence losing behavior prediction method based on enterprise relation map Download PDFInfo
- Publication number
- CN110991695A CN110991695A CN201911055174.7A CN201911055174A CN110991695A CN 110991695 A CN110991695 A CN 110991695A CN 201911055174 A CN201911055174 A CN 201911055174A CN 110991695 A CN110991695 A CN 110991695A
- Authority
- CN
- China
- Prior art keywords
- enterprise
- confidence
- risk
- losing
- relationship
- 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
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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Technology Law (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention provides a regional high-risk confidence losing behavior prediction method based on an enterprise relation map. It comprises the following steps: s1: establishing an enterprise relationship map according to the cooperation relationship among a plurality of enterprises; s2: predicting the high-risk districts losing confidence behaviors by using the enterprise relation map; s3: and calculating the likelihood value of the distrusted enterprise of each enterprise according to the preliminary prediction. The prediction method combines the enterprise relation map and the statistical data, so that the two parts of data can complement each other and cooperatively analyze the credit loss risk of a single enterprise and a single region, the credit loss behavior of the enterprise can be more comprehensively predicted, the prediction result is more visual, and the financial risk can be effectively reduced.
Description
Technical Field
The invention relates to the field of financial analysis, in particular to a regional high-risk confidence losing behavior prediction method based on an enterprise relation map.
Background
Financial network analysis can help people discover financial risks as early as possible, and the ability of resisting major risks is improved. The risk of loss of confidence is one of the important analytical aspects. However, the traditional financial analysis method mostly adopts data for division and does not relate to the contact relationship of enterprises. In addition, the enterprise relationship represented by the statistical data is not visual enough and is often difficult to locate quickly, and the analysis method is often difficult to determine the probability of losing confidence in the areas with complex enterprise relationship, so that the difficulty in predicting the financial risk is improved.
Disclosure of Invention
In order to solve the defects of the prior art, the invention aims to provide a regional high-risk confidence losing behavior prediction method based on an enterprise relationship map so as to solve the problem that the confidence losing probability is difficult to analyze in a region with a complex enterprise relationship.
In order to achieve the purpose, the invention provides a regional high-risk confidence losing behavior prediction method based on an enterprise relationship graph, which comprises the following steps: s1: establishing an enterprise relationship map according to the cooperation relationship among a plurality of enterprises; s2: predicting the high-risk districts losing confidence behaviors by using the enterprise relation map; s3: and calculating the likelihood value of the distrusted enterprise of each enterprise according to the preliminary prediction.
Preferably, in S1, the construction process of the enterprise relationship graph includes analyzing whether there is a loss of trust behavior of the enterprise, and marking the enterprise as black when there is a loss of trust behavior, and marking the enterprise as gray when there is no loss of trust behavior.
Preferably, in S1, a connected subgraph is displayed in the enterprise relationship graph, and the relationship between a distrusted enterprise and an un-distrusted enterprise is marked.
Preferably, in S1, the enterprise relationship graph is used to give the distribution of the number of all weakly connected sub-graph nodes.
Preferably, in S1, a plurality of smaller-scale connectivity graphs which can be directly drawn by a visualization method are selected from the enterprise relationship graph, so as to conveniently and intuitively see the structure of the investment network.
Preferably, in S2, the directional investment relationship network is converted into an undirected network for consideration, and the directionality of the connected edges is temporarily ignored.
Preferably, in S2, in the step S2, a probability P (B | m) of losing credit of the enterprise is defined, where P (B | m) represents a probability of losing credit of the enterprise with a number of losing credit greater than or equal to m in all neighbors.
Preferably, in S2, the probability of losing credit of the whole area is obtained by averaging the probability of losing credit of each enterprise in the area based on analyzing the probability of losing credit of the single enterprise.
Preferably, in S3, the single enterprise confidence loss likelihood value is obtained by using bernoulli distribution as a function family of the generalized linear model.
Compared with the prior art, the invention has the beneficial effects that:
(1) the invention combines the enterprise relation map and the statistical data, so that the two parts of data can complement each other and cooperatively analyze the credit loss risk of a single enterprise and area, can more comprehensively predict the credit loss behavior, has more intuitive prediction result and can effectively reduce the financial risk.
(2) The method utilizes Bernoulli distribution as a function family of the generalized linear model, analyzes the likelihood probability in the enterprise trust losing network, can depict the risk propagation of the financial security circle, and improves the accuracy of identifying key nodes in the financial security circle.
Drawings
FIG. 1 is a flow chart of a regional high risk loss of confidence behavior prediction method based on an enterprise relationship graph;
FIG. 2 is an enterprise relationship graph in a regional high risk loss of confidence behavior prediction method based on an enterprise relationship graph;
fig. 3 is a graph showing a variation curve of the enterprise loss probability in the regional high-risk loss behavior prediction method based on the enterprise relationship graph.
Detailed Description
To further understand the structure, characteristics and other objects of the present invention, the following detailed description is given with reference to the accompanying preferred embodiments, which are only used to illustrate the technical solutions of the present invention and are not to limit the present invention.
Fig. 1 is a flowchart of a regional high-risk confidence loss behavior prediction method based on an enterprise relationship graph. It comprises the following steps: s1: establishing an enterprise relationship map according to the cooperation relationship among a plurality of enterprises; s2: predicting the high-risk districts losing confidence behaviors by using the enterprise relation map; s3: and calculating the likelihood value of the distrusted enterprise of each enterprise according to the preliminary prediction.
Enterprises with the credit loss behavior are called credit loss enterprises for short in the patent.
In step S1, the process of constructing the enterprise relationship graph includes analyzing whether the enterprise has credit loss behavior, and marking the enterprise as black when the credit loss behavior exists and marking the enterprise as gray when the credit loss behavior does not exist.
And displaying a connected subgraph in the enterprise relation map, and marking the relation between the enterprise losing the information and the enterprise not losing the information.
Fig. 2 is an enterprise relationship diagram in a regional high risk loss of credit behavior prediction method based on an enterprise relationship diagram. And 4 representative weakly connected sub-graphs of different scales are displayed in the enterprise relationship graph, wherein gray nodes are enterprises without credibility behaviors, and black nodes are enterprises with credibility behaviors. The directions of the connecting edges in the map are not shown. Even if weak connectivity of the directed network is taken as a criterion, the network is not a fully connected network.
And giving the distribution of the number of all weakly connected sub-graph nodes by using the enterprise relation graph.
In the enterprise relationship graph in this example, only one super connected graph has a size of more than 100 ten thousand nodes, and the size of the rest second large connected graph has less than 10 ten thousand nodes. Most connectivity graphs are small in scale.
And selecting a plurality of connected graphs with smaller scale which can be directly drawn by a visualization method from the enterprise relation maps for conveniently and visually seeing the structure of the investment network.
The enterprise trust loss behavior has a significant "network effect" in the investment network between enterprises. That is, if the investor (the corporate shareholder) or the investment object (the subsidiary, the stock controlling company, the stock participating company, etc.) of a target enterprise has the loss behavior, the probability that the target enterprise has the loss behavior is also high.
In order to conveniently and intuitively observe the network effect, a simple situation is considered firstly, namely, a directed investment relation network is converted into an undirected network for consideration, and the directionality of a connecting edge is temporarily ignored.
In step S2, a probability P (B | m) of losing credit of the enterprise is defined, where P (B | m) represents the probability of losing credit of the enterprise with the number of losing credit being greater than or equal to m in all the neighbors.
When m is 0, P (B |0) is 0.0646, which is the proportion of the entire data set of the enterprise that is lost.
Fig. 3 is a graph showing a variation curve of the enterprise loss probability in the regional high-risk loss behavior prediction method based on the enterprise relationship graph. It describes the curve of P (B | m) as a function of m. The tendency of the curve to rise is very pronounced, especially when m is small, and when m is large, the data points are few, so that some fluctuation occurs.
From the test curve, it can be known that when m is 1, the probability that the target enterprise is a distrusted enterprise is sharply increased from P (B |0) 0.0646 to P (B |1) 0.1641.
If a business has 3 or more distressed businesses in its neighborhood, the probability of itself being a distressed business may exceed 40%.
As shown by the dashed lines, if the target enterprise is known to have no trusted enterprise in its neighborhood, the probability that the target enterprise is itself a trusted enterprise is 0.0474, which is 27% lower than the trusted enterprise in the entire data set.
In step S2, the probability of losing credit of the whole area is obtained by averaging the probability of losing credit of each enterprise in the area based on analyzing the probability of losing credit of the single enterprise through the variation graph of the probability of losing credit of the enterprise.
In addition, through the investment relationship network, some abnormal investment behaviors can be found, for example, through multiple changes, enterprise investment directed rings with the length s, such As enterprise A1 full asset share A2, A2 full asset share A3, …, As-1 full asset share As, As full asset share A1, are formed, and by utilizing the directed rings, enterprise A1 can register 1 hundred million yuan of capital to A2, A2 to A3, and the like, and finally returns to A1.
The capital investment is not real, but the real registered capital of each enterprise is increased by 1 billion yuan. These are all geigers that enterprises often use to acquire false qualifications, counterfeit project receiving capabilities, and even illegal funding. These significant financial economic risk problems translate into the classical graph theory problem of finding directed rings in a corporate investment relationship network of tens of millions of nodes.
The enterprise trust losing behavior also has obvious network effect, namely, if the credit losing behavior exists on the investor or the investment object of the target enterprise, the risk of the target enterprise for trust losing is greatly increased, and the trust losing risk is also rapidly increased along with the increase of the number of trust losing neighbors. This phenomenon may result from either risk transfer of the investment relationship itself (if the target enterprise's investment object has a credit breach indicating that the enterprise has a serious problem with cash flow, the target enterprise may have filled cash into the investment object and has a low probability of obtaining the desired investment income, and therefore the target enterprise itself may have similar risks), or from industry-wide problems (e.g., a large number of furniture industries are subject to administrative penalties after environmental standards have risen, and there are investment relationships in the industry chain among these same-industry enterprises).
In order to analyze the network effect of the enterprise credibility behavior, a generalized linear regression model is selected, the model fitting process is parallel, the calculation speed is very high, and the method is suitable for millions of enterprises to be processed in the text.
Because predicting the loss of confidence behavior of an enterprise is a typical binary problem, and bernoulli distribution is used as a function family of the generalized linear model, the likelihood value E (x, w) of an enterprise that is a loss of confidence enterprise can be fitted by the following calculation:
E(x,w)＝[1+exp(-wTx)]-1(1)
wherein x is a feature vector and w is a feature weight vector.
In step S3, the probability value of a single enterprise losing credit is obtained by using the bernoulli distribution as a function family of the generalized linear model so that one enterprise is the likelihood value of a losing credit enterprise.
It should be noted that the above summary and the detailed description are intended to demonstrate the practical application of the technical solutions provided by the present invention, and should not be construed as limiting the scope of the present invention. Various modifications, equivalent substitutions, or improvements may be made by those skilled in the art within the spirit and principles of the invention. The scope of the invention is to be determined by the appended claims.
Claims (9)
1. A regional high-risk confidence loss behavior prediction method based on an enterprise relationship graph is characterized by comprising the following steps: s1: establishing an enterprise relationship map according to the cooperation relationship among a plurality of enterprises; s2: predicting the high-risk districts losing confidence behaviors by using the enterprise relation map; s3: and calculating the likelihood value of the distrusted enterprise of each enterprise according to the preliminary prediction.
2. The method of claim 1, wherein in S1, the process of constructing the enterprise relationship graph includes analyzing whether there is an enterprise loss behavior, and marking the enterprise loss behavior as black when there is an enterprise loss behavior, and marking the enterprise loss behavior as gray when there is no enterprise loss behavior.
3. The method for regional high-risk confidence loss behavior prediction based on enterprise relationship graph of claim 1, wherein in the S1, a connected subgraph is displayed in the enterprise relationship graph and the relationship between the confidence losing enterprise and the non-confidence losing enterprise is marked.
4. The method for predicting regional high-risk distressing behavior based on enterprise relationship graph according to claim 1, wherein in S1, the enterprise relationship graph is used to give distribution of all weakly connected sub-graph nodes.
5. The method as claimed in claim 1, wherein in S1, a plurality of smaller-scale connectivity graphs which can be directly drawn by a visualization method are selected from the enterprise relationship graph, so as to conveniently and visually see the structure of the investment network.
6. The method of claim 1, wherein in step S2, the directed investment relationship network is converted into an undirected network for consideration, and the directionality of the connecting edge is temporarily ignored.
7. The method for forecasting regional high-risk credibility behaviors based on enterprise relationship graph of claim 1, wherein in step S2, in S2, an enterprise credibility probability P (B | m) is defined, where P (B | m) represents the credibility probability of enterprises with the number of credible enterprises in all neighbors greater than or equal to m.
8. The method for predicting high-risk districts losing confidence behaviors based on enterprise relationship maps of claim 1, wherein in S2, the probability of losing confidence of the whole district is obtained by averaging the probability of losing confidence of each enterprise in the district based on the analysis of the probability of losing confidence of each enterprise.
9. The method as claimed in claim 1, wherein in S3, the single enterprise confidence loss likelihood value is obtained by using bernoulli distribution as a function family of the generalized linear model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911055174.7A CN110991695A (en) | 2019-10-31 | 2019-10-31 | Regional high-risk confidence losing behavior prediction method based on enterprise relation map |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911055174.7A CN110991695A (en) | 2019-10-31 | 2019-10-31 | Regional high-risk confidence losing behavior prediction method based on enterprise relation map |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110991695A true CN110991695A (en) | 2020-04-10 |
Family
ID=70082762
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911055174.7A Pending CN110991695A (en) | 2019-10-31 | 2019-10-31 | Regional high-risk confidence losing behavior prediction method based on enterprise relation map |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110991695A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112348353A (en) * | 2020-11-05 | 2021-02-09 | 浪潮软件股份有限公司 | Enterprise confidence loss risk prediction method based on transfer learning |
-
2019
- 2019-10-31 CN CN201911055174.7A patent/CN110991695A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112348353A (en) * | 2020-11-05 | 2021-02-09 | 浪潮软件股份有限公司 | Enterprise confidence loss risk prediction method based on transfer learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Stock et al. | Twenty years of time series econometrics in ten pictures | |
| Anderson et al. | Some solutions to the multivariate Behrens–Fisher problem for dissimilarity‐based analyses | |
| CN101296128A (en) | Method for monitoring abnormal state of internet information | |
| Ge et al. | Detection of crossover time scales in multifractal detrended fluctuation analysis | |
| Kolody et al. | Evaluation of tag mixing assumptions in western Pacific Ocean skipjack tuna stock assessment models | |
| Liu et al. | An adaptive detection of multilevel co-location patterns based on natural neighborhoods | |
| Kalantari et al. | Spatio-temporal analysis of crime by developing a method to detect critical distances for the Knox test | |
| Braun et al. | Improving card fraud detection through suspicious pattern discovery | |
| Zhou et al. | The risk management using limit theory of statistics on extremes on the big data era | |
| CN111951104A (en) | Risk conduction early warning method based on associated graph | |
| CN110991695A (en) | Regional high-risk confidence losing behavior prediction method based on enterprise relation map | |
| Vochozka et al. | Model to predict survival of transportation and shipping companies | |
| James et al. | Equivalence relations and Lp distances between time series with application to the Black Summer Australian bushfires | |
| Liu et al. | An iterative detection and removal method for detecting spatial clusters of different densities | |
| Huva et al. | The impact of filtering self‐organizing maps: a case study with Australian pressure and rainfall | |
| Staňková et al. | On the influence of model setting on stochastic frontier analysis | |
| CN116383406A (en) | Enterprise portrait generation method, computer device and computer readable storage medium | |
| Tan et al. | Morphology‐based modeling of aggregation effect on the patch area size for GlobeLand30 data | |
| Manzhosov et al. | Method of constructing a visualization of threat model of information security | |
| CN110309313B (en) | Method and device for generating event transfer graph | |
| Kamal | A data mining approach for improving manufacturing processes quality control | |
| Darko et al. | The Chinese are Here: Firm Level Analysis of Import Competition and Performance in Sub-Saharan Africa | |
| Xia et al. | Characterizing the Outlying Feature Set of Groups | |
| CN111461199B (en) | Safety attribute selection method based on distributed junk mail classified data | |
| Withers et al. | Weighting cusums for increased power near the end points |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |