Process and system for developing a predictive model
First Claim
1. A computer implemented process for developing a model which predicts the value of a single dependent variable based on the value of at least one independent variable comprising the steps of:
- (a) providing a dataset containing a plurality of observations each containing a value for said dependent variable and values for said at least one independent variable;
(b) creating from said dataset of an initial generation of chromosomes each comprising a predictive model;
(c) determining a quantitative fitness measure for each chromosome in said initial generation; and
(d) creating a next generation of chromosomes by selecting a number of chromosomes from said initial generation, crossing said selected chromosomes by at least one of a cloning and a crossover technique, and mutating said chromosomes.
14 Assignments
0 Petitions
Accused Products
Abstract
The present invention relates to a computer implemented process for developing a model which predicts the value of a single dependent variable based on the value of at least one independent variable. The process comprises the steps of creating a dataset containing a plurality of observations each containing a value for the dependent variable and values for the at least one independent variable, creating from the dataset a plurality of original chromosomes each comprising a possible predictive model, developing a quantitative fitness measure for each chromosome, and creating a new generation of chromosomes by selecting a number of the original chromosomes based upon the fitness measures, crossing the selected original chromosomes by at least one of a cloning and a pure (standard) crossover technique, and mutating the crossed chromosomes. A system for carrying out the process of the present invention is also described.
16 Citations
48 Claims
-
1. A computer implemented process for developing a model which predicts the value of a single dependent variable based on the value of at least one independent variable comprising the steps of:
-
(a) providing a dataset containing a plurality of observations each containing a value for said dependent variable and values for said at least one independent variable;
(b) creating from said dataset of an initial generation of chromosomes each comprising a predictive model;
(c) determining a quantitative fitness measure for each chromosome in said initial generation; and
(d) creating a next generation of chromosomes by selecting a number of chromosomes from said initial generation, crossing said selected chromosomes by at least one of a cloning and a crossover technique, and mutating said chromosomes. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 35, 36, 37, 38, 39, 40)
-
-
34. A system for creating a predictive model comprising:
-
means for creating an initial generation of chromosomes from a dataset with each chromosome in said initial generation comprising a predictive model;
means for determining a quantitative fitness measure for each chromosome in said initial generation; and
means for selecting a number of said initial generation chromosomes, for crossing the selected initial generation chromosomes by at least one of cloning and a pure (standard) crossover technique, and for mutating the crossed chromosomes into a next generation of chromosomes.
-
- 41. A chromosome for predicting a model comprising a plurality of observed variable segments, a plurality of interaction segments and an intercept gene.
Specification