EP1261928A2

EP1261928A2 - System and method for valuing loan portfolios using fuzzy clustering

Info

Publication number: EP1261928A2
Application number: EP00982309A
Authority: EP
Inventors: Yu-To Chen; Christopher Donald Johnson; Tim Kerry Keyes; Chandrasekhar Pisupati; William Cree Steward
Original assignee: General Electric Co
Current assignee: General Electric Co
Priority date: 1999-12-02
Filing date: 2000-11-30
Publication date: 2002-12-04
Also published as: CN1630867A; PL358525A1; WO2001041016A8; BR0016142A; KR100686466B1; KR20020054362A; MXPA02005432A; WO2001041016A2; AU1935901A; JP2003526848A

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

SYSTEM AND METHOD FOR VALUING LOAN PORTFOLIOS USING FUZZY

CLUSTERING

CROSS REFERENCE TO RELATED APPLICATIONS

This application claims the benefit of U.S. Provisional Application Serial No. 60/168,499 filed on December 2, 1999, and entitled "System and Method for Valuing Commercial Loans Using Fuzzy Clustering and Knowledge Engineering for GECS Commercial Finance," which is incorporated by reference herein in its entirety.

BACKGROUND OF THE INVENTION

This disclosure relates to the valuation of loan portfolios, and more specifically, describes a system and method for valuating loan portfolios using clustering logic.

Generally, the valuation process involves determining the worth of income/cash flow generating assets. It is a relatively simple process that can be applied to expected streams of benefits from bonds, stocks, rental properties, oil v ells and loans. The valuation process is performed in order to determine the assets' worth at a give ^• point in time. The time value of money concept and the risk return concept are key elements in the valuation process. Quite simply, the valuation of any asset is equal to the net present value of all the expected future benefits, which are commonly measured in terms of cash flow.

Presently, it is difficult estimate the profitability valuation of an asset portfolio. Generally, an underwriter must go through loans in a portfolio one-by-one. This is very tedious and time consuming.

BRIEF SUMMARY OF THE INVENTION

This disclosure describes a system and method for the valuation of loan portfolios using clustering logic. Briefly described, in architecture, the system can be implemented as follows. An assets acquisition logic 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 profitability of the plurality of assets.

This disclosure can also be viewed as providing a method for valuating loan portfolios using clustering logic. In this regard, the method can be broadly summarized by the following steps: (1) acquiring a plurality of assets and attaching a plurality of variables to the assets; (2) examining the plurality of assets and variables with a model; (3) using fuzzy clustering to calculate the value of the plurality of assets and variables; and (4) calculating a profitability of the plurality of assets. These assets can be, but are not limited to, loan portfolios, bonds, stocks, rental property, etc.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram illustrating an example of the valuating of assets using a fuzzy clustering system with the example of loan portfolios.

FIG. 2 is a block diagram illustrating the valuation of assets using fuzzy clustering, situated w^'thin a computer readable medium in a computer system.

FIG. 3 is a block diagram illustrating an example of the process flow of the system and method for valuation of assets using fuzzy clustering.

FIG. 4 is a flow c art illustrating an example of the process flow of the system and method for evaluation of assets using fuzzy clustering, for an example of different types of loan portfolios.

FIG. 5 is a block diagram of an example illustrating the different types of data acquired during the data acquisition process, as shown in FIGs. 2, 3 and 4.

FIG. 6 is a table of examples of critical variables and the encoding schemes utilized in the hierarchical segmentation process, as shown in FIGs. 3 and 4. FIG. 7 is a block diagram of an example illustrating the use of segmentation modeling for the example of loan portfolios, as shown in FIGs. 2, 3 and 4 above.

FIG. 8 is a flow chart example of the fuzzy clustering process used in the system and method for valuation of assets of the present invention, for the example of loan portfolios, as shown in FIGs. 2, 3 and 4.

FIG. 9 is a flow chart of an example of the process that calculates the fuzzy clustering means in the system and method for asset valuation of the present invention, for the examples of loans and a portfolio, as shown in FIGs. 2, 3, 4 and 8.

FIG. 10 is a flow chart of an example of the underwriting review process in the system and method for valuing of assets of the present invention, for the example of loans in a portfolio, as shown in FIGs. 2, 3, and 4.

FIG. 1 1 A is a diagram illustrating an example of the intra-cluster variance of the six clusters generated in the dendrogram, as shown in FIG. 1 ID.

FIG. 1 IB is a diagram illustrating the inter-cluster variance of the six clusters, as shown in FIGs. 11A and 1 lC.

FIG. 11C is a diagram ref rred to as a dendrogram for illustrating each pair of cluster centroids using the distance matrix, as found in FIG. 8.

FIG. 1 ID is a diagram of a dendrogram generated during a rerun of the generate dendrogram, as shown in FIGs. 8 and 1 IB.

FIG. 12 is an example of a HELTR table of the valuation of the six clusters, as shown in FIG. 11D.

DETAILED DESCRIPTION OF THE INVENTION

Illustrated in FIG. 1 is a high-level view block diagram of the valuation of assets using the fuzzy clustering process. As shown, and discussed in more detail below, the process includes acquiring a number of portfolios of loans 2A-2Z. These portfolios of loans are then restructured using a portfolio restructuring process 3. The restructured portfolios are next fed into a merge loan information process 4, which takes loan information from a loan information database 5 and merges that loan information with the restructured portfolios. The restructured portfolios with the appended loan information 5 are then output as restructured portfolio 6A-6H. These restructured portfolios 6A-6H are then fed into a (expand) HELTR process 7 of the present invention. The HELTR process 7 provides a method for forming a valuation of assets using fuzzy clustering of the present invention. The valuing of assets using the fuzzy clustering method of the present invention takes the aforementioned loan data, then gives the expected cash flow and risk assessment for each of the restructured portfolios of assets 8.

As illustrated in FIG. 2, a computer system 21 generally comprises a processor 22 and memory 31 (e.g., RAM, ROM, hard disk, CD-ROM, etc.) with an operating system 32. The processor 22 accepts code and data from the memory 31 over the local interface 23, for example, a bus(es). Direction from the user can be signaled by using input devices, for example but not limited to, a mouse 24 and a keyboard 25.

The actions input and resulting output are displayed on the display terminal 26. An asset valuation using the fuzzy clustering system 50 can access other computers and resources on a network, utilizing modem or network card 27.

Also shown in FIG. 2 is a asset valuation using fuzzy clustering system 50 that includes: a data acquisition process 60, a variable selection process 80, a hierarchical segmentation process 100, a fuzzy clustering process 120, and an underwriting review process 180, which are in memory area 31. Databases 33 are also shown to reside in memory area 31. These components are herein described in further detail with regard to FIGS. 2-12. The memory area 31 can be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or propagation medium. More specific examples (a non-exhaustive list) of the memory area 31 include any one or more of the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM) (magnetic), a read-only memory (ROM) (magnetic), an erasable programmable read-only memory (EPROM or Flash memory) (magnetic), an optical fiber (optical), and a portable compact disc read-only memory (CDROM) (optical).

Illustrated in FIG. 3 is an example of the system and method flow for asset valuation using fuzzy clustering system 50 of this disclosure. The following description of the system and method for asset valuation using fuzzy clustering system 50 uses the example of loan portfolios. The assets can be, but are not limited to, loans, insurance policies, bonds, stocks, rental properties and other properties.

The system and method for asset valuation using fuzzy clustering system 50 includes the data acquisition process 60 that comprises the step of acquiring data. For the example used in the disclosure: loan portfolios, loan background information including loanee payment history data sets, credit analysis data sets, auto loan and mortgage loan data sets, and industry specific data sets. The data acquisition process 60 employs a divide and conquer approach to handling massive amounts of asset data. The asset data is input into the variable selection process 80 that identifies the critical loan variables for credit review or those variables with the most discriminating power and separating the various loan groups.

The loan portfolio data collected in the data acquisition process 60 and the critical variables identified in the variable selection process 80 are both input into the hierarchical segmentation process 100. The hierarchical segmentation process 100 segments the entire portfolio of assets (i.e., for this example, loans) into a number of bins based upon a predetermined critical variable selected by those reviewing the creditworthiness of the portfolios. After the hierarchical segmentation process 100 is performed, the segmented assets (i.e., loans) are further classified by the fuzzy clustering process 120. The fuzzy clustering process 120 classifies each of the segmented bins into a predetermined number of clusters based upon natural structure of the asset data. After this classification by the fuzzy clustering process 120 is performed, the classification of assets, (i.e., loans), are further processed by the underwriting review process 180 that assigns the projected cash flow and risk scores for the assets for each of the clusters. The projected cash flow and risk scores for the assets for each of the clusters is output for usage in the analysis of the creditworthiness of the portfolios. The data acquisition process 60, variable selection process 80, hierarchical segmentation process 100, the fuzzy clustering process 120 and underwriting review process 180 are hereindefined in further detail with regard to FIGs. 4-12.

Illustrated in FIG. 4 is a flow chart of the example of the asset valuation using fuzzy clustering system 50. First, the asset valuation using fuzzy clustering system 50 executes the data acquisition process 60 at step 51. The data acquisition process 60 is hereindefined in further detail with regard to FIG. 5.

The assets valuation using fuzzy clustering system 50 next executes the variable selection process 80 at step 52. The variable selection process 80 utilizes the asset data captured in the data acquisition process 60. The variable selection process 80 is hereindefined in further detail with regard to FIG. 6.

The asset valuation using fuzzy clustering system 50 then executes the hierarchical segmentation process 100. The hierarchical segmentation process 100 segments the portfolio of assets, i.e., loans, into a user predefined number of bins based upon the user identified critical variables. The hierarchical segmentation process 120 is hereindefined in further detail with regard to FIG. 7.

The asset valuation using fuzzy clustering system 50 next executes the fuzzy clustering process at step 54. The fuzzy clustering process 120 further classifies each of the bins identified in the hierarchical segmentation process 100 into a user predetermined number of clusters based upon the natural structure of the a^cset data. The fuzzy clustering process 120 is hereindefined in further detail with regard to FIG. 8.

The asset valuation of using fuzzy clustering system 50 executes the underwriting review process 180 at step 55. The underwriting review process 180 assigns the projected cash flow and risk scores to each of the clusters identified by the fuzzy clustering process 120. The underwriting review process 180 is hereindefined in further detail with regard to FIG. 10. At step 59, the asset valuation using fuzzy clustering system 50 then exits. Illustrated in FIG. 5 is a block diagram illustrating example types of asset databases utilized in building the attaching of asset relevant variables to the asset portfolio. The data acquisition process 60 includes the step of acquiring asset related data. This step generally comprises attaching data relevant to the assets in the asset portfolio. This disclosure will illustrate these concepts using a loan example discussed throughout this disclosure. Using this example, the loans will have relevant data cross-referenced and merged into the asset information. Preferably, the loan asset data in the loan asset database 60 are cross-referenced and merged with multiple databases.

For example, as shown in FIG. 5, the example loan asset database can include records from a variety of different universal files or databases including, for example, but not limited to, loanee payment history data sets 35, public credit analysis data sets 36, private credit analysis data sets 37, auto loan and mortgage loan data set data 38 and industry specific data sets 39. Preferably, the loan asset records in the loan assets database 60 are merged with the above-referenced data which will then be useful in identifying critical variables that are used during hierarchical segmentation process 100 and fuzzy clustering process 120 execution. Before inferring useful i ormation out of the composite date base, data scrubbing on the collected data is performed. For instance, data scru bing includes but is not limited to, detecting outliers, filing in or deleting missing values, derived imputed variables from raw data, etc.

Illustrated in FIG. 6 is a table of an example of one implementation of the variable selection process 80. In the variable selection process 80, the user identifies those variables that are deemed critical. In the present example, the variable selection process 80 has identified 11 variables that will be utilized by the fuzzy clustering process 120. As seen, associated with each of the variables deemed critical there is an associated category and/or value range for the variable as well as an encoding scheme for representing the variable.

Illustrated in FIG. 7 is an example of a hierarchical segmentation model created during the hierarchical segmentation process 100 with regard to the loan portfolios example. One example of a hierarchical segmentation model applied by the hierarchical segmentation process 100 is CART. CART is a well-known statistical algorithm of regressive trees and is used for hierarchical segmentation. The idea behind the regression trees is to segment the loan portfolios into a predetermined number of categories such that each category is homogenous with regard to the user predefined critical variables.

Shown in FIG. 7 is the result of the hierarchical segmentation process application of the CART model. The resulting regression tree has the example loan portfolios segmented using three critical variables. These three critical variables for this example include loan security, the loan type and the last payment on the loan. The resulting regression tree partitions the loan portfolio using these three critical variables into six bins. These partitions can be represented as a tree structure using the CART model. Once the loan portfolios are segmented by utilizing critical variables, the fuzzy clustering process 120 then can perform a fine granular partitioning of each of these predetermined bins.

Illustrated in FIG. 8 is a flow chart of an example one implementation of the fuzzy clustering process 120 of the present invention. The example implementation shown in FIG. 8 uses the example of the segmented loan portfolios generated by the hierarchical segmentation process 100.

The fuzzy clustering process 120 is first initialized at step 121. Next, at step 122, the fuzzy clustering process calculates the fuzzy clustering means by executing a calculate FCM process 140. The calculate FCM process 140 is hereindefined in further detail with regard to FIG. 9. At step 123, the fuzzy clustering process 120 calculates the intra-cluster and inter-cluster variance by box plot. This calculation of the intra-cluster and inter-cluster variance by box plot is a diagnostic check on the final result. This diagnostic check is performed by examining the corresponding box plots for the intra-cluster and inter-cluster, respectively. The intra-cluster and inter- cluster variance box-plots are hereindefined in further detail with regard to FIGs. 11 (A&B). At step 124, the fuzzy clustering process 120 determines if the intra-cluster and inter-cluster variance are compact enough or there is only one cluster left. Determining if the clustering is compact enough is resolved by determining if the intra-cluster variance is minimized while the inter-cluster variance is maximized by the calculate FMC process performed at step 122. If it is determined at step 124 that the clustering is compact enough or that only one is left, the fuzzy clustering process 120 exits at step 139.

If it is determined that the clustering is not compacted enough at step 124, the fuzzy clustering process 120 then gets the first pair on next set of cluster centroids at step 125. The fuzzy clustering process 120 calculates the distance between each of the pair of cluster centroids and stores this distance in a distance matrix at step 126. At step 131, the fuzzy clustering process 120 determines if there are more clustering centroid pairs to be examined. If there are more cluster centroid pairs to be examined, fuzzy clustering process 120 returns to repeat steps 125-131.

If there are no more pairs of cluster centroids to be processed, the fuzzy clustering process 120 then generates a dendrogram for each pair of cluster centroids using the distance matrix at step 132. The fuzzy clustering process 120 at step 133 inspects the dendrogram for possible merg es of cluster centroids. At step 133, the process inspects the dendrogram for possibu: merge of centroid clusters. The example of a dendrogram is hereindefined in further detail with regard to FIGs. 1 1 A through 11D.

At step 134, the fuzzy clustering process 120 determines if any cluster centroids can be merged. If there is possible merger of a pair of cluster centroids, the fuzzy clustering process 120 returns to repeat steps 122 through 124. If the fuzzy clustering process 120 determines that there is no possible merger of cluster centroids, the fuzzy clustering process 120 then exits at step 139. This process is herein defined in further detail with regard to FIGs. 1 1 A and 1 IB.

Illustrated in FIG. 9 is the calculate FCM process 140. First, in step 141, the number clusters and weighted exponents are input into the calculate FCM process 140. Next, the calculate FCM process 140 gets the first cluster at step 142. At step 143, the first (next) data point is obtained. The calculate FCM process 140 then randomizes the degree of membership of each point of each cluster at step 144. The degree of membership μ_lk is defined by

Intuitively, μ_lk, the degree of membership of the data point X_k in the cluster centroid V„ would get bigger as X_k is getting closer to V,. At the same time, μ_k would get smaller as X_k is getting farther away from V_j (other clusters). At step 145, the calculate FCM then determines whether all the data points in the current cluster have been randomized at step 145. If it is determined at step 145 that all of the data points have not been randomized, the calculate FCM process 140 then returns to repeat steps 143 through 145.

If the FCM process 140 does determine that all the data points for the current cluster have been randomized, the calculate FCM process 140 then determines if all the data points have been randomized for all of the available clusters at step 146. If the calculate FCM process 140 determines that not all the clusters have had their data points randomized, then the calculate FCM process 140 then returns to repeat steps 142 through 146.

If the calculate FCM process does determine that all the data pointsOf all the clusters have been randomized, the calculate FCM process 140 then calculates the centroid for all the data points at step 147. The rth cluster centroid V, is defined by

Intuitively, V„ is the ith cluster centroid, is the weighted sum of the coordinates of X_k, where k is the number of data points.

Starting with a desired number of clusters c and an initial guess for each cluster center V„ i = 1, 2,..., c, the calculate FCM process 140 will converge to a solution for V, that represents either a local minimum or a saddle point of the cost function. The quality of the calculate FCM process 140 solution, like that of most nonlinear optimization problems, depends strongly on the choice of initial values - the number c and the initial cluster centroids V,.

Next, at step 148, the calculate FCM process 140 calculates the objective function. The objective function is defined by

where n is the number of data points; c is the number of clusters, Xk is the kth data point; Vi is the z^'th cluster centroid; μ_k is the degree of membership of the kth data in the rth cluster; m is a constant greater than 1 (typically m=2). Note that μ_lk is a real number and bounded in [0, 1]. μ_lk = 1 means that ith data isdefinitely in kth cluster, while μ_lk = 0 means that ith data is definitely not in kth cluster. If μ_lk = 0.5, then it means that ith data is partially in kth cluster to the degree 0.5. Intuitively, the cost function would be minimized if each data point belongs exactly to a specific cluster and there is no partial degree of membership to any other clusters. That is, there is no ambiguity in assigning each data point to the cluster where it belongs.

At step 149, the calculate FCM process 140 determines whether the value of the calculate objective function is convergent. If the calculate objective function is not convergent, the calculate FCM process 140 proceeds to steps 151 through 155. If the calculate FCM process 140 determines that the value of the objective function is convergent at step 149, the calculate FCM process then exits at step 159.

At step 151, the calculate FCM process 140 gets the first cluster. At step 152, the first data point is obtained. The FCM process 140 then updates the degree of membership of each point in each cluster at step 153. At step 154, the calculate FCM process 140 determines whether every data point in the current cluster has been updated. If all the data points in the current cluster have not been updated, the calculate FCM process 140 then returns to repeat steps 152 through 154. If all the data points for the current cluster have been updated, the calculate FCM process 140 next determines whether each data point of all the clusters has been updated at step 155.

If the calculate FCM process 140 determines that not all data points have been updated for all clusters at step 155, the calculate FCM process 140 then returns to repeat steps 151 through 155. If the calculate FCM process 140 determines that all the data points for all the clusters have been updated at step 155, the calculate FCM process 140 then returns to repeat steps 147-149, as defined above.

Illustrated in FIG. 10 is underwriting review process 180. The underwriting review process 180 is performed after the entire portfolio is segmented by the fuzzy clustering processing 120. During the underwriting review process 180, each cluster is reviewed and is assigned a composite score called HELTR. HELTR s Ηnds for: H- high cash flow; E-expected cash flow; L-low cash flow; T-timing of cash flow in months; and R-risk assessment of borrower. In essence, the HELTR score captures both the expected and range of cash, the timing of cash flow and the risk associate 1 with each cluster.

First, the underwriting review process 180 is initialized at step 181. At step

182, the first loan segment is obtained and made the current loan segment. At step

183, the first cluster in the current loan segment is obtained. At step 184, the underwriting review process 180 calculates the cash flow score and timing of the cash flow for the current cluster in the current loan segment. At step 185, the underwriting review process 180 calculates the risk assessment of purchasing all the clusters in the current loan segment. At step 186, the underwriting review process determines if the assessment of all the current clusters in the current loan segment have been performed. If there are more clusters in the current loan segment, the underwriting review process 180 returns to repeat step 183 through 186. At step 187, the underwriting review process 180 determines whether all the clusters have been reviewed in all of the loan segments If the underwriting review process 180 determines that not all the clusters in all the loan segments have been reviewed, the underwriting review process 180 returns to repeat steps 182 through 187 If the underwriting review process 180 determines that all of the clusters in all of the loan segments have been reviewed, the underwriting review process 180 then exits at step 189

Illustrated in FIG 1 1A and 1 IB are the intra-cluster and inter-cluster variance of an example of six clusters as generated at step 123 (FIG 8) As shown in FIG 1 IB, the average distance of all the data points to centroid one is 1 0, while the average distance of the data points in cluster 1 is 0 6 as shown in FIG 1 1A This indicates that the clusteπng is quite compact Therefore, the intra-cluster variance is minimized, while the inter-cluster variance is maximized by the calculate FCM process 140

Illustrated in FIGs 1 1C and 1 ID are dendrograms of 20 clusters and 6 c'usters, respectively. It is noted that each dendrogram starts with a data point forming a separate cluster, and in which data points or clusters close to one another are succes^sfully merged It usually requires a couple iterations of the fuzzy clustering process 12 to get the optimum of amount of clusters As shown in FIG 11C, of the 20 centroids, 12 of them should be merged Therefore, in the above example, the fu^zy clusteπng process 120 is then re-executed with six clusters with the result being shown in FIG 11D.

Illustrated in FIG. 12 is an example of a HELTR table for the six clusters shown in FIGs. 1 ID As shown, the HELTR table includes data for each centroid identified in the example in FIG 7 and for each cluster within each identified centroid The data utilized in the HELTR table include whether or not the loans are secured, revolving credit, required notice, include a last payment, loan maturity, or whether or not the loans are guaranteed Shown are the collective score, en position, unpaid pπnciple balance in millions, total unpaid principal %, and cash flow analysis The method and system for the valuation of assets using fuzzy clustering system 50 comprise an ordered listing of executable instructions for implementing logical functions. The ordered listing can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device, and execute the instructions. In the context of this document, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

The computer readable medium can be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or propagation medium. More specific examples (a nonexhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette

(magnetic), a random access memory (RAM) (magnetic), a read-only memory (ROM) (magnetic), an erasable programmable read-only memory (EPROM or Flash memory) (magnetic), an optical fiber (optical), and a portable compact disc read-only memory (CDROM) (optical).

Note that the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.

The foregoing description has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Obvious modifications or variations are possible in light of the above teachings. The flow charts of this disclosure show the architecture, functionality, and operation of a possible implementation of the register usage optimization compilation and translation system. In this regard, each block represents a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the blocks may occur out of the order noted in the figures, or for example, may in fact be executed substantially concurrently or in the reverse order, depending upon the functionality involved.

The system and methods discussed were chosen and described to provide the best illustration of the principles of the invention and its practical application to enable one of ordinary skill in the art to utilize the invention in various embodiments and with various modifications as are suited to the particular use contemplated. All such modifications and variations are within the scope of the invention as determined by the appended claims when interpreted in accordance with the breadth to which they are fairly and legally entitled.

Claims

WHAT IS CLAIMED IS:

1. A method for providing asset valuation using fuzzy clustering comprising the steps of:

acquiring a plurality of assets and attaching a plurality of variables to said plurality of assets;

examining said plurality of assets and said plurality of variables with a model;

using fuzzy clustering to calculate the value of said plurality of assets and said variables; and

calculating a profitability of said plurality of assets.

2. The method of claim 1, wherein said step of using fuzzy clustering further comprises the step of:

randomizing a degree of membership of each of a plurality of data points in each of a plurality of clusters; and

calculating a centroid for said plurality of data points and said plurality of clusters.

3. The method of claim 2, wherein said step of using fuzzy clustering further comprises the step of:

updating a degree of membership of each of said plurality of data points in each of said plurality of clusters.

4. The method of claim 3, wherein said step of using fuzzy clustering further comprises the step of:

calculating a value with an objective function; and

determining if said objective function value is convergent.

5. The method of claim 2, wherein said step of using fuzzy clustering further comprises the steps of:

calculating inter-cluster variants; and

calculating intra-cluster variants.

6. The method of claim 5, wherein said step of using fuzzy clustering further comprises the step of:

compacting said inter-cluster variants and said intra-cluster variants.

7. The method of claim 6, wherein sa step of compacting said inter- cluster variants and said intra-cluster variants further comprises the steps of:

calculating the distance between each pair of "aid inter-cluster variants and said intra-cluster variants;

storing a distance between each pair of said inter-cluster variants and said intra-cluster variants in distance matrix;

generating a dendrogram for said distance between each pair of said inter-cluster variants and said intra-cluster; and

evaluating the dendrogram for possible mergers.

8. A system for providing asset valuation using fuzzy clustering comprising:

means for acquiring a plurality of assets;

means for attaching a plurality of variables to said plurality of assets;

means for examining said plurality of assets and said plurality of variables with a model;

means for using fuzzy clustering to calculate the value of said plurality of assets and said variables; and

means for calculating a profitability of said plurality of assets.

9. The system of claim 8, wherein said using fuzzy clustering means further comprises:

means for randomizing a degree of membership of each of a plurality of data points in each of a plurality of clusters; and

means for calculating a centroid for said plurality of data points and said plurality of clusters.

10. The system of claim 9, wherein said using fuzzy clustering means further comprises:

means for updating a degree of membership of each of said plurality of data points in each of said plurality of clusters.

11. The system of claim 10, wherein said using fuzzy clustering means further comprises:

means for calculating a value with an objective function; and

means for determining if said value is convergent.

12. The system of claim 8, wherein said using fuzzy clustering means further comprises:

means for calculating inter-cluster variants; and

means for calculating intra-cluster variants.

13. The system of claim 12, wherein said using fuzzy clustering means further comprises:

means for compacting said inter-cluster variants and said intra-cluster variants.

14. The system of claim 13, wherein said using fuzzy clustering means further comprises:

means for calculating the distance between each pair of said inter- cluster variants and said intra-cluster variants;

means fo. storing a distance between each pair of said inter-cluster variants and said intra-cluster variants in distance matrix;

means for generating dendrogram for said distance between each pair of said inter-cluster variants and said intra-cluster; and

means for evaluating the dendrogram for possible mergers.

15. A system for providing stability analysis of profitability for different types of goods or services products comprising:

data acquisition logic that acquires a plurality of assets and attaches a plurality of variables to said plurality of assets;

segmentation logic that examines said plurality of assets and said plurality of variables with a model;

fuzzy clustering logic that calculates a value of said plurality of assets and said variables; and

profitability analysis logic that calculates a profitability of said plurality of assets.

16. The system of claim 15, wherein said fuzzy clustering logic further comprises:

logic that randomizes a degree of membership of each of a plurality of data points in each of a plurality of clusters; and

logic that calculatts a centroid for said plurality of data points and said plurality of clusters.

17. The system of claim 16, further comprises:

logic that updates a degree of membership of each of said plurality of data points in each of said plurality of clusters.

18. The system of claim 17, further comprising:

logic for calculating a value with an objective function and determining if said value is convergent.

19. The system of claim 17, further comprising:

logic that calculates inter-cluster variants from said plurality of data points in each of said plurality of clusters; and

logic that calculates intra-cluster variants from said plurality of data points in each of said plurality of clusters.

20. The system of claim 19, further comprising:

logic that compacts said inter-cluster variants and said intra-cluster variants.

21. The system of claim 16, further comprising:

logic calculating a distance between each pair of said inter-cluster variants and said intra-cluster variants;

logic storing a distance between each pair of said inter-cluster variants and said intra-cluster variants in a distance matrix;

logic generating dendrogram f ~>r said distance between each pair of said inter-cluster variants and said intra-cluste ^•; and

logic evaluating the dendrogram for possible mergers.