CN1630867A - System and method for valuing loan portfolios using fuzzy clustering - Google Patents
System and method for valuing loan portfolios using fuzzy clustering Download PDFInfo
- Publication number
- CN1630867A CN1630867A CNA008187681A CN00818768A CN1630867A CN 1630867 A CN1630867 A CN 1630867A CN A008187681 A CNA008187681 A CN A008187681A CN 00818768 A CN00818768 A CN 00818768A CN 1630867 A CN1630867 A CN 1630867A
- Authority
- CN
- China
- Prior art keywords
- cluster
- assets
- fuzzy clustering
- logic
- calculate
- 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/02—Banking, e.g. interest calculation or account maintenance
-
- 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
- G06Q40/06—Asset management; Financial planning or analysis
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Technology Law (AREA)
- Theoretical Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Human Resources & Organizations (AREA)
- Operations Research (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
This disclosure describes a system and method for the valuation of loan portfolios using clustering logic. The system of this disclosure includes an assets acquisition logic that acquires a plurality of assets and attaches a plurality of variables to the plurality of assets in an assets database. Segmentation logic examines the plurality of assets and variables with a model. A fuzzy clustering logic calculates a value of the plurality of assets and variables, and a profitability analysis logic calculates a profilitability of the plurality of assets. The method valuates loan portfolios using clustering logic. The method includes the steps of acquiring a plurality of assets and attaching a plurality of variables to the assets. The plurality of assets and variables are examined with a model using fuzzy clustering to calculate the value of the plurality of assets and variables. The profitability of the plurality of assets is then calculated.
Description
The cross reference of related application
This application requires to enjoy U. S. application 60/168 that propose on Dec 2nd, 1999, " System andMethod for Valuing Commercial Loans Using Fuzzy Clustering andKnowledge Engineering for GECS Commercial Finance " by name, 499 right of priority, the full content of this application in this is comprised in as a reference.
Background technology of the present invention
This openly relates to the appraisal of loan assets combination, and more particularly, this has openly been described one and has been used to use the cluster logic loan assets to be made up the system and method for making appraisal.
Usually, appraisal process relates to the assets value of determining that income/cash flow produces.It is a simple relatively process, can be applied to the expectation stream of acquiring an advantage from bond, stock, leased property, oil vells and loan.Carrying out this evaluation procedure is for the value the definite assets of some preset time.It is key element in this evaluation procedure that the timevalue concept of money and risk are given back notion.Very briefly, the appraisal of any assets equals the net present value (NPV) of the future benefit of all expectations, and it is generally measured according to cash flow.
At present, the yield rate appraisal of assets of expectation is difficult.Usually, the consignee's debt-credit of scrutiny in an asset portfolio one by one.This is very dull and time-consuming.
Summary of the invention
This has openly described a kind of system and method that is used to use the combination of cluster logic appraisal loan assets.Concise and to the point, in architecture, this system can be implemented as follows.Assets obtain that logic is obtained a plurality of assets and in an asset database additional a plurality of variablees to these a plurality of assets.Cut apart logic with a pattern checking this a plurality of assets and variable.A value of a plurality of assets of fuzzy clustering logical calculated and variable, and yield rate analysis logic calculates the yield rate of a plurality of assets.
This openly can also be regarded as providing a kind of method that is used to use the combination of cluster logic appraisal loan assets.In this, this method can roughly be summarised as the following step: (1) obtains a plurality of assets, and additional a plurality of variablees are on these assets; (2) use a kind of pattern checking this a plurality of assets and variable; (3) use fuzzy clustering to calculate the value of a plurality of assets and variable; And (4) calculate the yield rate of these a plurality of assets.These assets can be but be not limited to, loan assets combination, bond, stock, leased property etc.
Brief Description Of Drawings
Fig. 1 is a block diagram, with the loan assets example combinations example that uses fuzzy clustering system appraisal assets has been described.
Fig. 2 is a block diagram, has illustrated to use the appraisal of fuzzy clustering to assets, and it is located in computer-readable medium in the computer system.
Fig. 3 is a block diagram, and an example of the treatment scheme of the system and method that is used to use fuzzy clustering appraisal assets has been described.
Fig. 4 is a process flow diagram, and an example of the treatment scheme of the system and method that is used to use fuzzy clustering appraisal assets has been described with regard to a dissimilar loan assets example combinations.
Fig. 5 is a block diagram that an example of the different types of data that obtains during the data acquisition shown in Fig. 2,3 and 4 has been described.
Fig. 6 is the key variables that use in the cutting procedure of a layering shown in Fig. 3 and 4 and a form of encoding scheme example.
Fig. 7 be one illustrated shown in above Fig. 2,3 and 4, be used for the example block diagram of use of the parted pattern of loan assets example combinations.
Fig. 8 is used for evaluating an example flow diagram of the fuzzy clustering flow process that the system and method for assets uses with regard to the loan assets example combinations, shown in Fig. 2,3 and 4, in the present invention.
Fig. 9 is the example of just loan and asset portfolio, the process flow diagram of an example of system and method calculating fuzzy clustering method flow process shown in Fig. 2,3,4 and 8, be used for the valuation of assets in the present invention.
Figure 10 is with regard to the loan example in the asset portfolio, process flow diagram shown in Fig. 2,3 and 4, check an example of flow process consigning of the system and method that is used for the valuation of assets of the present invention.
Figure 11 A is a block diagram, and an example of difference in the cluster of six clusters that produce in the dendrogram shown in Figure 11 D has been described.
Figure 11 B is a block diagram, and the cluster differences of six clusters shown in Figure 11 A and the 11C has been described.
Figure 11 C is a block diagram that is called as dendrogram, is used for the service range matrix each centre of moment to cluster is described, as finding among Fig. 8.
Figure 11 D is the block diagram of a dendrogram that produces during the reruning of this generation dendrogram shown in Fig. 8 and 11B.
Figure 12 is the example of a HELTR table of these six the cluster appraisals shown in Figure 11 D.
Detailed description of the present invention
Fig. 1 has illustrated the high-level view block diagram that uses the fuzzy clustering process to carry out the valuation of assets.As shown in the figure, and discussed in detail below such, and this process comprises: obtain a plurality of loan assets combination 2A-2Z.Use an asset portfolio restructuring procedure 3 to re-construct these loan assets combinations then.Next this asset portfolio that re-constructs is admitted to a consolidation loan information process 4, and this process is obtained credit information from a credit information database 5, and that credit information and this asset portfolio that re-constructs are merged.Has asset portfolio additional facility information 5, that re-construct then as the asset portfolio 6A-6H output that re-constructs.These asset portfolio 6A-6H that re-construct are admitted to (expansion) HELTR process 7 of the present invention then.HELTR process 7 provides a kind of fuzzy clustering that is used for using the present invention to form the method for the valuation of assets.Use the fuzzy clustering method among the present invention to carry out the valuation of assets, utilize above-mentioned loan data, provide the cash flow and the evaluation of risk of expectation then for the asset portfolio 8 of each reconstruct.
As shown in Figure 2, computer system 21 storer 31 (for example, RAM, ROM, hard disk, CD-ROM, etc.) of comprising a processor 22 usually and having an operating system 32.Processor 22 is via local interface 23, for example (many) bus receives code and data from storer 31.By the use input equipment, such as, but be not limited to, mouse 24 and keyboard 25 just can be by signalings from user's indication.Action input and result's output are displayed on the display terminal 26.One is used the valuation of assets system 50 of fuzzy clustering can utilize modulator-demodular unit or other computing machine and the resource of network interface card 27 visits on a network.
Also shown a system 50 that uses fuzzy clustering appraisal assets among Fig. 2, it comprises: the cutting procedure of data acquisition process 60, Variables Selection process 80, a layering 100, a fuzzy clustering process 120 and the checking process 180 of consigning, these processes are all in storage area 31.Also shown the database 33 that resides in the storage area 31.These assemblies are described in more detail at this with regard to Fig. 2-12.This storage area 31 can be, such as, but be not limited to, electronics, magnetic, optics, electromagnetism, infrared ray or semiconductor system, device, equipment or a propagation medium.More instantiations of storage area 31 (not exhaustive tabulation) comprise in following any one or a plurality of: the electrical connection (electronics) with one or more electric wire, a portable computer diskette (magnetic), a random-access memory (ram) (magnetic), a ROM (read-only memory) (ROM) (magnetic), an EPROM (Erasable Programmable Read Only Memory) (EPROM or flash memory) (magnetic), an optical fiber (optics), an and portable optic disk ROM (read-only memory) (CDROM) (optics).
An example of the system and method flow process that is used for the system 50 that this is disclosed, assets are evaluated in the use fuzzy clustering has been described among Fig. 3.Be used to use the following description utilization of a loan asset portfolio example of system and method for the system 50 of fuzzy clustering appraisal assets.These assets can be, but be not limited to loan, insurance slip, bond, stock, leased property and other property.
The system and method that is used to use fuzzy clustering to evaluate the system 50 of assets comprises data acquisition 60, and this process comprises the step of obtaining data.The example that uses in the disclosure has: loan assets makes up, and includes the loan background information of debtor's payment history data acquisition, credit analysis data acquisition, loan for purchasing car and mortgage loan data acquisition and the concrete data acquisition of industry.This data acquisition 60 uses the method that solves that distributes to handle a large amount of asset datas.This asset data is imported in the Variables Selection process 80, and this process identifier is used for key loan variable or those variablees that has maximum resolving ability and separate various loan groups of credit inspection.
Loan assets data splitting of collecting in data acquisition 60 and the key variable that identifies in Variables Selection process 80 all are imported in the cutting procedure 100 of layering.Based on predetermined key variables of selecting by the worth credit degree of checking asset portfolio, the cutting procedure 100 of this layering is assigned to a plurality of bins to the assets of whole asset portfolio (that is the loan that, is used for this example).After the cutting procedure 100 of having carried out this layering, these assets of cutting apart (that is loan) are further classified by fuzzy clustering process 120.Based on the natural structure of this asset data, fuzzy clustering process 120 is categorized into each section in the cluster of a predetermined quantity.After carrying out this classification by this fuzzy clustering process 120, assets (that is further handled by the checking process 180 of consigning, and this process is the cash flow and the value-at-risk of the asset allocation planning of each cluster by) classification, loan.Output is used for the planning cash flow and the value-at-risk of each cluster assets, is used for using the credit standing analysis of this asset portfolio.With reference to figure 4-12, the here cutting procedure 100 of definition of data acquisition process 60, Variables Selection process 80, layering, fuzzy clustering process 120 and the checking process 180 of consigning in more detail.
A process flow diagram of the example of the system 50 that uses fuzzy clustering appraisal assets has been described among Fig. 4.At first, the valuation of assets system 50 with fuzzy clustering carries out data acquisition 60 in step 51.Defined data acquisition 60 with reference to figure 5 in more detail at this.
The system 50 that uses fuzzy clustering appraisal assets is next in step 52 performance variable selection course 80.This Variables Selection process 80 is used the asset data that obtains in data acquisition 60.Defined Variables Selection process 80 with reference to figure 6 in more detail at this.
The system 50 that uses fuzzy clustering to evaluate assets carries out the cutting procedure 100 of layering then.The cutting procedure 100 of this layering is based on the key variables of user ID, assets, be that loan portfolio is assigned in the bin of a consumer premise justice quantity.Defined the cutting procedure 100 of layering here in more detail with reference to figure 7.
Use the system 50 of fuzzy clustering appraisal assets next to carry out the fuzzy clustering process in step 54.This fuzzy clustering process 120 further is categorized into each bin of sign in the cutting procedure 100 of layering in the cluster of a consumer premise quantity based on the natural structure of asset data.Defined fuzzy clustering process 120 in more detail with reference to figure 8 here.
Use the system 50 of fuzzy clustering appraisal assets next to carry out the checking process 180 of consigning in step 55.This checking process 180 of consigning is distributed planning cash flow and relative risk value for each by the cluster of these fuzzy clustering process 120 signs.With reference to Figure 10 the checking process 180 of consigning is carried out further specific definition here.In step 59, use the system 50 of fuzzy clustering appraisal assets to withdraw from then.
A block diagram has been described among Fig. 5, this block diagram illustrations the exemplary types of the asset database that uses in the asset portfolio at the relevant variable of the additional assets of structure.This data acquisition 60 comprises the step of obtaining the assets related data.This step comprises usually: the additional data relevant with the assets in this asset portfolio.This openly will use a loan example of always discussing in this is open that these notions are described.Use this example, loan will make related data by cross reference and be merged in the assets information.Preferably be that the loan assets data in this loan assets database 60 merge by cross reference and with a plurality of databases.
For example, as shown in Figure 5, this example loan assets database can comprise the record from various different general files or database, these general files or database comprise, such as, but be not limited to, debtor's payment history data acquisition 35, open credit analysis data acquisition 36, private credit analyze data acquisition 37, loan for purchasing car and mortgage loan data acquisition data 38 and the concrete data acquisition 39 of industry.Preferably be, loan assets record in this loan assets database 60 merges with the above-mentioned data of quoting, these data of quoting will be useful in the identification key variables then, and wherein this key variable is in the cutting procedure 100 of layering and the term of execution use of fuzzy clustering process 120.Before from the database that is synthesized, inferring Useful Information, carry out the data erase on gathering information.For example, data erase including but not limited to, detect outlier, input or deletion missing value, from raw data, derive estimated value, etc.
A form realizing example of Variables Selection process 80 has been described among Fig. 6.In this Variables Selection process 80, those are considered to crucial variable User Recognition.In this example, this Variables Selection process 80 has been discerned 11 variablees that will be used by fuzzy clustering process 120.Can see a related category and/or a value scope that is used for this variable and an encoding scheme that is used to represent this variable being arranged with each is considered to that crucial variable is associated.
The example of the parted pattern of a layering of creating with regard to this loan assets combination examples, during the cutting procedure 100 of this layering has been described in Fig. 7.A layering examples of segmentation models of being used by the cutting procedure 100 of this layering is CART.CART is a kind of well-known regression tree statistic algorithm, and is used for layering and cuts apart.In this regression tree thought behind is that the loan assets combination is divided in the classification of a predetermined number, and therefore each classification all is an isomorphism for the key variables of this consumer premise justice.
The result who has shown the layering cutting procedure application of CART model among Fig. 7.The regression tree that is produced has the example loan assets combination of using three key variables to cut apart.These three key variables that are used for this example comprise loan security, loan types and the payment last time on this loan.The regression tree that produces uses these three key variables that this loan assets combination is divided in six bins.These divisions can be represented as a tree structure that uses the CART model.In case the combination of this loan assets is by using after key variables have been cut apart, fuzzy clustering process 120 can be carried out in these predetermined bins the particulate of each then and divide.
A process flow diagram realizing example of fuzzy clustering process 120 of the present invention has been described among Fig. 8.The realization example that shows in Fig. 8 uses by what this layering cutting procedure 100 produced cuts apart the loan assets combination examples.
In step 124, fuzzy clustering process 120 determine the inner and cluster differences of these clusters whether be enough compactness or whether only have a cluster to stay.By determining whether to be minimized and whether the cluster differences is maximized, can determine that whether this cluster is enough compactness by the cluster internal diversity that the calculating FMC process of carrying out in step 122 obtains.If determine that in step 124 cluster is enough compact or only has only one to be left, then fuzzy clustering process 120 withdraws from step 139.
If determine that in step 124 this cluster is not enough compact, then fuzzy clustering process 120 is step 125 acquisition the first couple on next group cluster centre of moment.Fuzzy clustering process 120 is calculated at each in step 126 and is stored in the distance matrix to the distance between the cluster centre of moment and this distance.In step 131, fuzzy clustering process 120 has determined whether that the more cluster centre of moment is to being examined.If the more cluster centre of moment is arranged to being examined, then fuzzy clustering process 120 is returned with repeating step 125-131.
If do not have the more cluster centre of moment to processed, it is that each produces a dendrogram to the cluster centre of moment that then fuzzy clustering process 120 is used this distance matrix in step 132.Fuzzy clustering process 120 is checked this dendrogram in step 133 for may merging of the cluster centre of moment.In step 133, this process is may merging of centre of moment cluster to check this dendrogram.To 11D a dendrogram example is carried out further specific definition with reference to figure 11A here.
In step 134, fuzzy clustering process 120 determined whether any can the merged cluster centre of moment.The merging of a pair of cluster centre of moment is if possible arranged, and then fuzzy clustering process 120 is returned with repeating step 122 to 124.If fuzzy clustering process 120 determines there be not may merging of the cluster centre of moment, then fuzzy clustering process 120 withdraws from step 139.With reference to figure 11A and 11B this process is carried out further detailed definition here.
Calculating FCM process 140 has been described among Fig. 9.At first, in step 141, number of clusters and weighted index are imported into and calculate in the FCM process 140.Next, calculate FCM process 140 and obtain first cluster in step 142.In step 143, obtain first (next one) data point.Calculate the degree of membership of FCM process 140 then at each point of each cluster of step 144 randomization.This degree of membership μ
IkBe defined as
See intuitively, at cluster centre of moment V
iData point X
kDegree of membership μ
IkWill be along with X
kThe closer to V
iAnd become big more.Simultaneously, along with X
kMore away from V
j(other cluster) μ
IkIt is more little to become.In step 145, calculate FCM and determine whether all data points in current cluster have been randomized then.If determine that in step 145 not all data point all has been randomized, then calculate the FCM process and return with repeating step 143 to 145.
If FCM process 140 determines all data points of current clusters really and all be randomized, then this calculating FCM process 140 determines whether that for all available clusters all data points all have been randomized in step 146.Determine that not every cluster has all allowed their data point be randomized if calculate FCM process 140, then calculate FCM process 140 and return with repeating step 142 to 146.
If this calculating FCM process is determined all data points of all clusters really and all has been randomized, then calculates FCM process 140 is calculated all data points in step 147 the centre of moment.This i cluster centre of moment V
iBe defined as
Intuitively as can be seen, i cluster centre of moment V
iBe X
kThe weighted sum of coordinate, wherein k is the number of data point.
From a clusters number c who wants be used for each cluster centre of moment V
i, i=1,2 ..., c, an initial estimation begin, this calculating FCM process 140 will converge to one and be used for V
iSeparate, its expression cost function a local minimum or a saddle point.The quality that this calculating FCM process 140 is separated as most of nonlinear optimal problem, seriously depends on initial value-quantity c and initial clustering centre of moment V
iSelection.
Next, in step 148, calculate FCM process 140 calculating target functions.This objective function is defined as:
Wherein n is the number of data point; C is a clusters number, X
kBe k data point; V
iBe i the cluster centre of moment; μ
IkIt is the degree of membership of k data in i cluster; M be one greater than 1 constant (general m=2).Notice μ
IkBe a real number and be limited in [0,1].μ
Ik=1 means i data clearly in k cluster, and μ
Ik=0 means that i data are clear and definite not in k cluster.If μ
Ik=0.5, this means that then the degree of i data part in k cluster is 0.5.Intuitively, if each data point belongs to a concrete cluster just and do not take office the what part degree of membership of its cluster, then this cost function will be minimized.In other words, in distributing the cluster of each data point under, there is not ambiguity to it.
In step 149, calculate FCM process 140 and determine whether the value of calculating target function is convergent.If the objective function that calculates is not a convergent, then calculates FCM process 140 and proceed step 151 to 155.Determine that in step 149 value of this objective function is a convergent if calculate FCM process 140, then calculate the FCM process and withdraw from step 159.
In step 151, calculate FCM process 140 and obtain first cluster.In step 152, obtain first data point.Calculate FCM process 140 then and be updated in the degree of membership of each point in each cluster in step 153.In step 154, calculate FCM process 140 and determine whether each data point in current cluster has been updated.If not all data point all has been updated in current cluster, then calculates FCM process 140 and return with repeating step 152 to 154.All be updated if be used for all data points of current cluster, then calculated FCM process 140 and next determine whether that in step 155 each data point of all clusters all has been updated.
Determine it is not all to have upgraded all data points if calculate FCM process 140 in step 155, then calculate FCM process 140 and return with repeating step 151 to 155 for all clusters.Determine all to have upgraded all data points in step 155 if calculate FCM process 140, then calculate FCM process 140 and return with repeating step 147-149, as defined above for all clusters.
The checking process 180 of consigning has been described among Figure 10.This checking process 180 of consigning is carried out after whole asset portfolio is cut apart by this fuzzy clustering process 120.During the checking process 180 of consigning, each cluster all has been examined, and has been assigned with an integrate score that is known as HELTR.HELTR representative: the high cash flow of H-; E-expects cash flow; L-hangs down cash flow; T-is the cash flow timing of unit with the moon; And R-borrower's risk assessment.In fact, the HELTR call cash and the cash scope of expectation, the cash flow timing and with the relative risk of each cluster correlation.
At first, in the step 181 initialization checking process 180 of consigning.In step 182, obtain first loan splitting, and be used as current loan splitting.In step 183, obtain first cluster in current loan splitting.In step 184, this checking process 180 of consigning is current cluster calculation cash flow score and cash flow timing in current loan splitting.In step 185, the checking process of consigning 180 is calculated the risk assessment of buying all clusters in current loan splitting.In step 186, the checking process of consigning determines whether the assessment of all clusters in current loan splitting all has been performed.If more cluster is arranged in current loan splitting, then this checking process 180 of consigning is returned with repeating step 183 to 186.
In step 187, the checking process of consigning 180 determines whether all clusters in all loan splittings all have been examined.Checking process 180 determines it is not that all clusters in all loan splittings all have been examined if this is consigned, and then this checking process 180 of consigning is returned with repeating step 182 to 187.Checking process 180 determines that all clusters in all loan splittings all have been examined if this is consigned, and then this checking process 180 of consigning withdraws from step 189.
Inner and the cluster differences of cluster of six cluster examples that produce in step 123 (Fig. 8) has been described among Figure 11 A and the 11B.Shown in Figure 11 B, all data points are 1.0 to the mean distance of the centre of moment 1, and the mean distance as the data point in Figure 11 cluster that A is shown in 1 is 0.6 simultaneously.This represents that this cluster is very compact.Therefore, by calculating FCM process 140, the cluster internal diversity has been minimized, and the cluster differences has been maximized simultaneously.
The dendrogram of 20 clusters and 6 clusters has been described respectively among Figure 11 C and the 11D.It should be noted that: all since a data point that forms an independent cluster, wherein approximating data point or cluster have successfully been merged each dendrogram.It needs the coupling of fuzzy clustering process 120 usually and repeats to obtain the optimal number of cluster.Shown in Figure 11 C, in the middle of 20 centre of moments, there are 12 should be merged.Therefore, in above-mentioned example, re-execute fuzzy clustering process 120 with six clusters, its result shows in Figure 11 D.
A HELTR form example that is used for six clusters that Figure 11 D shows has been described among Figure 12.As shown in the figure, this HELTR form comprises and is used in the data of each centre of moment that Fig. 7 example is discerned and the data that are used for each cluster in each identification centre of moment.The data of using in this HELTR form comprise: this loan whether be maintain secrecy, revolver, need notice, comprise a nearest payment, whether provide a loan date of expiry or this loan have been assured.Having shown the score of collecting, the right of retention position, is the unpaid basic balance of unit, unpaid capital % and fund flow analysis fully with 1,000,000.
The method and system that is used to use fuzzy clustering to evaluate the system 50 of assets comprises the ordered list of executable instruction that is used to realize logic function.This ordered list can be comprised in the computer-readable medium, be used for using by an instruction execution system, device or equipment, perhaps be used in combination with them, these systems, device or equipment be such as a computer based system, comprise the system of processor or other can from this instruction execution system, device or taking-up instruction and carry out the system of this instruction.In the environment of this document, " computer-readable medium " can be any device that is used for the program being used or used in conjunction with their by this instruction execution system, device or equipment that can comprise, stores, transmits, propagates or transmit.
Computer-readable medium can be, such as, but be not limited to, electronics, magnetic, optics, electromagnetism, infrared ray or semiconductor system, device, equipment or a propagation medium.More instantiations of this computer-readable medium (not exhaustive tabulation) will comprise following: electrical connection (electronics), a portable computer diskette (magnetic), a random-access memory (ram) (magnetic), a ROM (read-only memory) (ROM) (magnetic), an erasable programmable ordered magnetism (EPROM or flash memory) (magnetic), an optical fiber (optics) and the portable optical magnetic (CD-ROM) (optics) with one or more electric wire.
Notice, this computer-readable medium in addition can be paper or other print the suitable medium of this program in the above, because if necessary words, this program can via, for example, the optical scanning of this paper or other medium is obtained electronically, then by compiling, explanation, perhaps opposite processed in a kind of suitable mode, be stored in then in the computer memory.
For the purpose of illustration and description has provided foregoing description.It is not attempted exhaustive or limits the invention to disclosed precise forms.According to above-mentioned teaching, obviously may make amendment or change.This disclosed process flow diagram has shown that this register uses a structure in the cards, function and the operation of optimizing compiling and translation system.In this respect, can represent module, section or a partial code for every, it comprises the logic function (a plurality of function) that one or more executable instruction is used to realize appointment.Should also be noted that in some replace to be realized, depend on the function that relates to, the function of in this piece, indicating can be not with in the figure shown in the identical order of order occur, perhaps for example, can carry out in fact simultaneously or carry out with inverted order.
The system and method for being discussed is selected and describe so that the best illustration to the principle of the invention and its practical application to be provided, so that make those of ordinary skills can use the present invention in various embodiments, and the various modifications that are suitable for the specific use considered are arranged.When being explained by amplitude just and the legitimate claim protection according to them, all this modifications and variations are all within the scope of the invention of determining by accessory claim.
Claims (21)
1. one kind is used to provide the method for using fuzzy clustering appraisal assets, comprises step:
Obtain a plurality of assets and additional a plurality of variablees to described a plurality of assets;
With the described a plurality of assets of pattern checking and described a plurality of variable;
Use fuzzy clustering to calculate the value of described a plurality of assets and described variable; And
Calculate a yield rate of described a plurality of assets.
2. the method for claim 1, it is characterized in that: the step of described use fuzzy clustering further comprises following steps:
Randomization is in each degree of membership of a plurality of data points of a plurality of clusters in each; And
Be described a plurality of data points and centre of moment of described a plurality of cluster calculation.
3. method as claimed in claim 2 is characterized in that: the step of described use fuzzy clustering further comprises following steps:
Be updated in described a plurality of data points of described a plurality of cluster in each degree of membership of each.
4. method as claimed in claim 3 is characterized in that: the step of described use fuzzy clustering further comprises following steps:
Calculate a value with an objective function; And
Determine whether described target function value restrains.
5. method as claimed in claim 2 is characterized in that: the step of described use fuzzy clustering further comprises following steps:
Calculate the difference between the cluster; And
Calculate the difference of cluster inside.
6. method as claimed in claim 5 is characterized in that: the step of described use fuzzy clustering further comprises following steps:
Compress described cluster differences and described cluster internal diversity.
7. method as claimed in claim 6 is characterized in that: the step of described cluster differences of described compression and described cluster internal diversity further comprises following steps:
Calculating at each to the distance between described cluster differences and the described cluster internal diversity;
The distance between described cluster differences and the described cluster internal diversity being stored in the distance matrix at each;
For the distance between described cluster differences and the described cluster internal diversity being produced a dendrogram at each; And
Estimation is used for the dendrogram that may merge.
8. one kind is used to provide the system that uses fuzzy clustering appraisal assets, comprises:
Be used to obtain the device of a plurality of assets;
Be used for the device of additional a plurality of variablees to described a plurality of assets;
Be used to use the device of the described a plurality of assets of pattern checking and described a plurality of variablees;
Be used to use fuzzy clustering to calculate the device of the value of described a plurality of assets and described variable; And
Be used to calculate the device of a yield rate of described a plurality of assets.
9. system as claimed in claim 8 is characterized in that: the device of described use fuzzy clustering further comprises:
The device that is used for randomization degree of membership of each in each a plurality of data points of a plurality of clusters; And
Be used to the device of the described a plurality of data point and a centre of moment of described a plurality of cluster calculation.
10. system as claimed in claim 9 is characterized in that: the device of described use fuzzy clustering further comprises:
Be used for being updated in the device of the degree of membership of each in each described a plurality of data points of described a plurality of cluster.
11. system as claimed in claim 10 is characterized in that: the device of described use fuzzy clustering further comprises:
Be used to use an objective function to calculate the device of a value; And
Be used for determining whether convergent device of described target function value.
12. system as claimed in claim 8 is characterized in that: the device of described use fuzzy clustering further comprises:
Be used to calculate the device of cluster differences; And
Be used to calculate the device of cluster internal diversity.
13. system as claimed in claim 12 is characterized in that: the device of described use fuzzy clustering further comprises:
Be used to compress the device of described cluster differences and described cluster internal diversity.
14. system as claimed in claim 13 is characterized in that: the device of described use fuzzy clustering further comprises:
Be used to calculate at each device to the distance between described cluster differences and the described cluster internal diversity;
Be used for the distance between described cluster differences and the described cluster internal diversity being stored into the device of distance matrix at each;
Be used to the device that the distance between described cluster differences and the described cluster internal diversity is produced a dendrogram at each; And
Be used to estimate the device that is used for the dendrogram that may merge.
15. a system that is used to dissimilar commodity or service product that the yield rate stability analysis is provided comprises:
Data are obtained logic, obtain a plurality of assets and additional a plurality of variablees to described a plurality of assets;
Cut apart logic, with the described a plurality of assets of pattern checking and described a plurality of variable;
The fuzzy clustering logic is calculated the value of described a plurality of assets and described variable; And
The yield rate analysis logic calculates a yield rate of described a plurality of assets.
16. system as claimed in claim 15 is characterized in that: described fuzzy clustering logic further comprises:
The logic that is used for randomization degree of membership of each in each a plurality of data points of a plurality of clusters; And
Logic for the described a plurality of data points and a centre of moment of described a plurality of cluster calculation.
17. system as claimed in claim 16 further comprises:
Be used for being updated in the logic of the degree of membership of each in each described a plurality of data points of described a plurality of cluster.
18. system as claimed in claim 17 further comprises:
Be used to use objective function to calculate whether convergent logic of a value and definite described value.
19. system as claimed in claim 17 further comprises:
From in the described a plurality of data points of described a plurality of clusters each, calculating the logic of cluster differences; And
From in the described a plurality of data points of described a plurality of clusters each, calculating the logic of cluster internal diversity.
20. system as claimed in claim 19 further comprises:
Compress the logic of described cluster differences and described cluster internal diversity.
21. system as claimed in claim 16 further comprises:
Calculating is in each logic to the distance between described cluster differences and the described cluster internal diversity;
The distance between described cluster differences and the described cluster internal diversity being stored into logic in the distance matrix at each;
For the distance between described cluster differences and the described cluster internal diversity being produced the logic of a dendrogram at each; And
Estimation is used for the logic of the dendrogram that may merge.
Applications Claiming Priority (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US16849999P | 1999-12-02 | 1999-12-02 | |
| US60/168,499 | 1999-12-02 | ||
| US56188600A | 2000-05-01 | 2000-05-01 | |
| US09/561,886 | 2000-05-01 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1630867A true CN1630867A (en) | 2005-06-22 |
Family
ID=26864184
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA008187681A Pending CN1630867A (en) | 1999-12-02 | 2000-11-30 | System and method for valuing loan portfolios using fuzzy clustering |
Country Status (9)
| Country | Link |
|---|---|
| EP (1) | EP1261928A2 (en) |
| JP (1) | JP2003526848A (en) |
| KR (1) | KR100686466B1 (en) |
| CN (1) | CN1630867A (en) |
| AU (1) | AU1935901A (en) |
| BR (1) | BR0016142A (en) |
| MX (1) | MXPA02005432A (en) |
| PL (1) | PL358525A1 (en) |
| WO (1) | WO2001041016A2 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105528730A (en) * | 2015-12-15 | 2016-04-27 | 杜衡 | Asset pool targeting algorithm based on asset securitization |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130238395A1 (en) * | 2010-11-27 | 2013-09-12 | Jerry Z. Shan | Composite Driver Derivation |
| WO2018028666A1 (en) * | 2016-08-12 | 2018-02-15 | 正大天晴药业集团股份有限公司 | Crystal of dpp-iv long-acting inhibitor and salt thereof |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3555211B2 (en) * | 1994-07-06 | 2004-08-18 | オムロン株式会社 | Database search apparatus and method, direct mail issuance support system equipped with database search apparatus |
| JP3334029B2 (en) * | 1996-03-28 | 2002-10-15 | 日本電信電話株式会社 | Cluster classification method and cluster classification device |
| JPH10275177A (en) * | 1997-03-28 | 1998-10-13 | Nri & Ncc Co Ltd | Device and method for evaluating performance of investment trust |
-
2000
- 2000-11-30 CN CNA008187681A patent/CN1630867A/en active Pending
- 2000-11-30 BR BR0016142-0A patent/BR0016142A/en not_active Application Discontinuation
- 2000-11-30 EP EP00982309A patent/EP1261928A2/en not_active Withdrawn
- 2000-11-30 WO PCT/US2000/032592 patent/WO2001041016A2/en not_active Application Discontinuation
- 2000-11-30 PL PL00358525A patent/PL358525A1/en not_active Application Discontinuation
- 2000-11-30 JP JP2001542001A patent/JP2003526848A/en active Pending
- 2000-11-30 MX MXPA02005432A patent/MXPA02005432A/en not_active Application Discontinuation
- 2000-11-30 AU AU19359/01A patent/AU1935901A/en not_active Abandoned
- 2000-11-30 KR KR1020027007056A patent/KR100686466B1/en not_active IP Right Cessation
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105528730A (en) * | 2015-12-15 | 2016-04-27 | 杜衡 | Asset pool targeting algorithm based on asset securitization |
| CN105528730B (en) * | 2015-12-15 | 2021-11-09 | 杜衡 | Asset pool targeting method based on asset securitization |
Also Published As
| Publication number | Publication date |
|---|---|
| MXPA02005432A (en) | 2002-11-29 |
| KR100686466B1 (en) | 2007-02-23 |
| EP1261928A2 (en) | 2002-12-04 |
| BR0016142A (en) | 2002-12-24 |
| WO2001041016A2 (en) | 2001-06-07 |
| KR20020054362A (en) | 2002-07-06 |
| AU1935901A (en) | 2001-06-12 |
| JP2003526848A (en) | 2003-09-09 |
| WO2001041016A8 (en) | 2002-02-07 |
| PL358525A1 (en) | 2004-08-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Houston et al. | The financial implications of supply chain changes | |
| Wang et al. | Use of data envelopment analysis in assessing information technology impact on firm performance | |
| CN108665159A (en) | A kind of methods of risk assessment, device, terminal device and storage medium | |
| CN108416669A (en) | User behavior data processing method, device, electronic equipment and computer-readable medium | |
| WO2007106786A2 (en) | Methods and systems for multi-credit reporting agency data modeling | |
| US20040225473A1 (en) | Fraud score calculating program, method of calculating fraud score, and fraud score calculating system for credit cards | |
| CN104850965A (en) | Intelligent financial system financial data automatic generation method and intelligent financial system | |
| CN107437227A (en) | Stock investment analysis apparatus and method | |
| Tsolas | Firm credit risk evaluation: a series two-stage DEA modeling framework | |
| CN112990386B (en) | User value clustering method and device, computer equipment and storage medium | |
| CN111552680A (en) | Construction method and device of business fraud identification database and computer equipment | |
| CN112861980B (en) | Calendar task table mining method based on big data and computer equipment | |
| CN112950347B (en) | Resource data processing optimization method and device, storage medium and terminal | |
| Cao et al. | Bond rating using support vector machine | |
| Kumar et al. | Forecasting credit ratings Using ANN and statistical techniques | |
| CN115170869A (en) | Repeated vehicle damage claim identification method, device, equipment and storage medium | |
| Olson | Data mining in business services | |
| WO2007073063A1 (en) | Method of technology valuation | |
| CN111932368B (en) | Credit card issuing system and construction method and device thereof | |
| CN1630867A (en) | System and method for valuing loan portfolios using fuzzy clustering | |
| CN116843483A (en) | Vehicle insurance claim settlement method, device, computer equipment and storage medium | |
| CN116402625A (en) | Customer evaluation method, apparatus, computer device and storage medium | |
| CN116451125A (en) | New energy vehicle owner identification method, device, equipment and storage medium | |
| CN110245823A (en) | Shops's distribution method and system based on overdue number of days | |
| CN113706258A (en) | Product recommendation method, device, equipment and storage medium based on combined model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |