Methods and systems for modeling using classification and regression trees

• US 7,120,599 B2
• Filed: 12/21/2000
• Issued: 10/10/2006
• Est. Priority Date: 12/30/1999
• Status: Active Grant

• Current Assignee: General Electric Company
• Original Assignee: GE Capital Commercial Finance Incorporated (General Electric Company)
• Inventors: Keyes, Tim Kerry
• Primary Examiner(s): Hamilton, Lalita M.
• Application Number:
  US09/746,411
• Publication Number:
  US 20010044766A1
• Time in Patent Office: 2,119 Days
• Field of Search: 705/35, 705/36, 705/37, 705/36.R
• US Class Current: 705/36.00R
• CPC Class Codes: G06Q 40/00 Finance; Insurance; Tax strategies; Processing of corporate or income taxes
• CPC Class Codes: G06Q 40/06 Asset management; Financial planning or analysis

Abstract

A method of valuation of large groups of assets using classification and regression trees is described. The method includes defining relevant portfolio segmentations, assessing performance of the classification and regression tree based model against a simple model and ranking all portfolio segments based upon performance of the models. Iterative and adaptive statistical evaluation of all assets and statistical inferences are used to generate the segmentations. The assets are collected into a database, grouped by credit variable, subdivided by ratings as to those variables and then rated individually. The assets are then regrouped and a collective valuation is established by cumulating individual valuations.

Claims

1. A computer implemented method for grouping assets included within a portfolio of assets for valuation purposes using a classification and regression tree based model, said method comprising the steps of:

receiving from a seller a proposal to sell a portfolio of assets, the assets included within at least one segment defined by the seller;

computing sum of squared error (SSE) values for the at least one defined portfolio segment using the classification and regression tree based model and the computer, wherein the classification and regression tree based model generates at least one cluster of assets included within the portfolio of assets for valuing each non-underwritten asset included within the at least one cluster, each non-underwritten asset included within the at least one cluster is assigned a value based on an average value assigned to underwritten assets included within the at least one cluster;

computing SSE values for the at least one defined portfolio segment using a simple model and the computer, wherein the simple model assigns a single value to all non-underwritten assets within the at least one defined portfolio segment based on a value assigned to at least one underwritten asset included within the at least one defined portfolio segment;

computing an error ratio between the SSE values based on the classification and regression tree based model and the SSE values based on the simple model for the at least one defined portfolio segment using the computer;

ranking the at least one defined portfolio segment based on the computed error ratio; and

using the ranking by a potential buyer to determine an amount to offer for purchasing assets included within the at least one defined portfolio segment.

2. A method according to claim 1 wherein said step of receiving from a seller a proposal to sell a portfolio of assets, the assets included within at least one segment defined by the seller, further comprises receiving at least one defined portfolio segment that is defined by the seller based upon at least one of pre-defined tranches, unpaid balance amounts, region customer risk and loan rankings.

3. A method according to claim 1 further comprising the step of using the classification and regression tree based model as a predictor for the at least one defined portfolio segment if the computed error ratio for at least one defined portfolio segment is less than one.

4. A method according to claim 1 wherein the at least one defined portfolio segment is a plurality of defined portfolio segments, said method further comprising computing coefficient of determination (R²) values for each asset within each defined portfolio segment, where R² per asset is computed as (sum of squares total (SST) per segment−

SSE per segment)/(overall SST for all assets within the portfolio×

number of assets within each segment).

5. A method according to claim 4 further comprising ranking portfolio segments of the plurality of defined portfolio segments based on the computed R² values.

6. A method according to claim 4 further comprising:

using at least one of the classification and regression tree based model and the simple model as a predictor for a first portfolio segment of the plurality of defined portfolio segments; and

using a computed R² value of a second portfolio segment of the plurality of defined portfolio segments as a predictor for the second portfolio segment.

7. A method according to claim 1 further comprising ranking the at least one defined portfolio segment based upon a coefficient of determination (R²) value computed for the at least one defined portfolio segment, where R²=1−

(SSE/sum of squares total (SST)).

8. A method according to claim 1 wherein the at least one defined portfolio segment is a plurality of defined portfolio segments, said method further comprising:

using the classification and regression tree based model as a predictor for a first portfolio segment of the plurality of defined portfolio segments; and

using the simple model as a predictor for a second portfolio segment of the plurality of defined portfolio segments.

9. A method according to claim 1 wherein the simple model assigns a single value to all non-underwritten assets within the at least one defined portfolio segment based on at least one of an average value, a median value, a highest value, and a lowest value of the underwritten assets within the at least one defined portfolio segment.

10. A method according to claim 1 wherein the simple model assigns a single value to all non-underwritten assets within the portfolio of assets based on a value of an underwritten asset within the portfolio of assets.

11. A system for grouping assets included within a portfolio of assets for valuation purposes using a classification and regression tree based model, said system comprising:

a computer configured as a server and a database of asset portfolios; and

at least one client system connected to said server through a network, said server configured to;

receive from a seller a proposal to sell a portfolio of assets, the assets included within at least one segment defined by the seller;

compute sum of squared error (SSE) values for the at least one defined portfolio segment using the classification and regression tree based model, wherein the classification and regression tree based model generates at least one cluster of assets included within the portfolio of assets for valuing each non-underwritten asset included with the at least one cluster, each non-underwritten asset included within the at least one cluster is assigned a value based on an average value assigned to underwritten assets included within the at least one cluster;

compute SSE values for the at least one defined portfolio segment using a simple model, wherein the simple model assigns a single value to all non-underwritten assets within the at least one defined portfolio segment based on a value assigned to at least one underwritten asset included within the at least one defined portfolio segment;

compute an error ratio between the SSE values based on the classification and regression tree based model and the SSE values based on the simple model for the at least one defined portfolio segment;

rank the at least one defined portfolio segment based on the computed error ratio; and

determine an amount to offer by a potential buyer for purchasing assets included within the at least one defined portfolio segment based on the ranking.

12. A system according to claim 11 wherein the at least one defined portfolio segment is defined by the seller based upon at least one of pre-defined tranches, unpaid balance amounts, region customer risk and asset rankings.

13. A system according to claim 11 wherein said server is configured to use the classification and regression tree based model as a predictor for the at least one defined portfolio segment if the computed error ratio for the at least one defined portfolio segment is less than one.

14. A system according to claim 11 wherein the at least one defined portfolio segment is a plurality of defined portfolio segments, said server configured to compute determination of coefficient (R²) values for each asset within each defined portfolio segment, where R² per asset is computed as (sum of squares total (SST) per segment−

SSE per segment)/(overall SST for all assets×

number of assets within each segment).

15. A system according to claim 14 wherein said server is further configured to rank portfolio segments of the plurality of defined portfolio segments based on the computed R² values.

16. A system according to claim 14 wherein said server is further configured to:

use at least one of the classification and regression tree based model and the simple model as a predictor for a first portfolio segment of the plurality of defined portfolio segments; and

use a computed R² value of a second portfolio segment of the plurality of defined portfolio segments as a predictor for the second portfolio segment.

17. A system according to claim 11 wherein said server is configured to rank the at least one defined portfolio segment based upon a coefficient of determination (R²) value computed for the at least one defined portfolio segment, where R²=1−

(SSE/sum of squares total (SST)).

18. A system according to claim 11 wherein the at least one defined portfolio segment is a plurality of defined portfolio segments, said server further configured to:

use the classification and regression tree based model as a predictor for a first portfolio segment of the plurality of defined portfolio segments; and

use the simple model as a predictor for a second portfolio segment of the plurality of defined portfolio segments.

19. A system according to claim 11 wherein the simple model assigns a single value to all non-underwritten assets within the at least one defined portfolio segment based on at least one of an average value, a median value, a highest value, and a lowest value of the underwritten assets within the at least one defined portfolio segment.

20. A system according to claim 11 wherein the simple model assigns a single value to all non-underwritten assets within the portfolio of assets based on a value of an underwritten asset within the portfolio of assets.

21. A computer configured for grouping assets included within a portfolio of assets for valuation purposes using a classification and regression tree based model, said computer including a database of portfolios of assets, said computer programmed to:

receive from a seller a proposal to sell a portfolio of assets, the assets included within at least one segment defined by the seller;

compute sum of squared error (SSE) values for the at least one defined portfolio segment using the classification and regression tree based model, wherein the classification and regression tree based model generates at least one cluster of assets included within the portfolio of assets for valuing each non-underwritten asset included within the at least one cluster, each non-underwritten asset included within the at least one cluster is assigned a value based on an average value assigned to underwritten assets included within the at least one cluster;

compute SSE values for the at least one defined portfolio segment using a simple model, wherein the simple model assigns a single value to all non-underwritten assets within the at least one defined portfolio segment based on a value assigned to at least one underwritten asset included within the at least one defined portfolio segment;

compute an error ratio between the SSE values based on the classification and regression tree based model and the SSE values based on the simple model for the at least one defined portfolio segment;

rank the at least one defined portfolio segment based on the computed error ratio; and

determine an amount to offer by a potential buyer for purchasing assets included within the at least one defined portfolio segment based on the ranking.

22. A computer according to claim 21 wherein the at least one defined portfolio segment is defined by the seller based upon at least one of pre-defined tranches, unpaid balance amounts, region customer risk and asset rankings.

23. A computer according to claim 21 programmed to use the classification and regression tree based model as a predictor for the at least one defined portfolio segment if the computed error ratio for the at least one defined portfolio segment is less than one.

24. A computer according to claim 21 wherein the at least one defined portfolio segment is a plurality of defined portfolio segments, said computer programmed to compute determination of coefficient (R²) values for each asset within each defined portfolio segment, where R² per asset is computed as (sum of squares total (SST) per segment−

SSE per segment)/(overall SST for all assets×

number of assets within each segment).

25. A computer according to claim 24 wherein said computer is further programmed to rank portfolio segments of the plurality of defined portfolio segments based on the computed R² values.

26. A computer according to claim 24 wherein said computer is further programmed to:

use at least one of the classification and regression tree based model and the simple model as a predictor for a first portfolio segment of the plurality of defined portfolio segments; and

use a computed R² value of a second portfolio segment of the plurality of defined portfolio segments as a predictor for the second portfolio segment.

27. A computer according to claim 21 programmed to rank the at least one defined portfolio segment based upon a coefficient of determination (R²) value for the at least one defined portfolio segment, where R²=1−

(SSE/sum of squares total (SST)).

28. A computer according to claim 21 wherein the at least one defined portfolio segment is a plurality of defined portfolio segments, said computer further programmed to:

use the classification and regression tree based model as a predictor for a first portfolio segment of the plurality of defined portfolio segments; and

use the simple model as a predictor for a second portfolio segment of the plurality of defined portfolio segments.

29. A computer according to claim 21 wherein the simple model assigns a single value to all non-underwritten assets within the at least one defined portfolio segment based on at least one of an average value, a median value, a highest value, and a lowest value of the underwritten assets within the at least one defined portfolio segment.

30. A computer according to claim 21 wherein the simple model assigns a single value to all non-underwritten assets within the portfolio of assets based on a value of an underwritten asset within the portfolio of assets.