Misclassification cost minimizing fitness functions for genetic algorithm-based artificial neural network classifiers

Research output: Contribution to journalArticle

7 Citations (Scopus)

Abstract

We study three different approaches to formulate a misclassification cost minimizing genetic algorithm (GA) fitness function for a GA-neural network classifier. These three different approaches include a fitness function that directly minimizes total misclassification cost, a fitness function that uses posterior probability for minimizing total misclassification cost and a hybrid fitness function that uses an average value of the first two fitness functions to minimize total misclassification cost. Using simulated data sets representing three different distributions and four different misclassification cost matrices, we test the performance of the three fitness functions on a two-group classification problem. Our results indicate that the posterior probability-based misclassification cost minimizing function and the hybrid fitness function are less prone to training data over fitting, but direct misclassification cost minimizing fitness function provides the lowest overall misclassification cost in training tests. For holdout sample tests, when cost asymmetries are low (less than or equal to a ratio of 1:2), the hybrid misclassification cost minimizing fitness function yields the best results; however, when cost asymmetries are high (equal or greater than a ratio of 1:4), the total misclassification cost minimizing function provides the best results. We validate our findings using a real-world data on a bankruptcy prediction problem.Journal of the Operational Research Society (2009) 60, 1123-1134. doi:10.1057/palgrave.jors.2602641; published online 25 June 2008.

Original languageEnglish (US)
Pages (from-to)1123-1134
Number of pages12
JournalJournal of the Operational Research Society
Volume60
Issue number8
DOIs
StatePublished - Aug 1 2009

Fingerprint

Classifiers
Genetic algorithms
Neural networks
Costs
Cost functions
Misclassification
Fitness
Classifier
Artificial neural network
Genetic algorithm

All Science Journal Classification (ASJC) codes

  • Management Information Systems
  • Strategy and Management
  • Management Science and Operations Research
  • Marketing

Cite this

@article{86b0f7cfea13468096cf78e0fe7f5cdd,
title = "Misclassification cost minimizing fitness functions for genetic algorithm-based artificial neural network classifiers",
abstract = "We study three different approaches to formulate a misclassification cost minimizing genetic algorithm (GA) fitness function for a GA-neural network classifier. These three different approaches include a fitness function that directly minimizes total misclassification cost, a fitness function that uses posterior probability for minimizing total misclassification cost and a hybrid fitness function that uses an average value of the first two fitness functions to minimize total misclassification cost. Using simulated data sets representing three different distributions and four different misclassification cost matrices, we test the performance of the three fitness functions on a two-group classification problem. Our results indicate that the posterior probability-based misclassification cost minimizing function and the hybrid fitness function are less prone to training data over fitting, but direct misclassification cost minimizing fitness function provides the lowest overall misclassification cost in training tests. For holdout sample tests, when cost asymmetries are low (less than or equal to a ratio of 1:2), the hybrid misclassification cost minimizing fitness function yields the best results; however, when cost asymmetries are high (equal or greater than a ratio of 1:4), the total misclassification cost minimizing function provides the best results. We validate our findings using a real-world data on a bankruptcy prediction problem.Journal of the Operational Research Society (2009) 60, 1123-1134. doi:10.1057/palgrave.jors.2602641; published online 25 June 2008.",
author = "Pendharkar, {Parag C.}",
year = "2009",
month = "8",
day = "1",
doi = "10.1057/palgrave.jors.2602641",
language = "English (US)",
volume = "60",
pages = "1123--1134",
journal = "Journal of the Operational Research Society",
issn = "0160-5682",
publisher = "Palgrave Macmillan Ltd.",
number = "8",

}

TY - JOUR

T1 - Misclassification cost minimizing fitness functions for genetic algorithm-based artificial neural network classifiers

AU - Pendharkar, Parag C.

PY - 2009/8/1

Y1 - 2009/8/1

N2 - We study three different approaches to formulate a misclassification cost minimizing genetic algorithm (GA) fitness function for a GA-neural network classifier. These three different approaches include a fitness function that directly minimizes total misclassification cost, a fitness function that uses posterior probability for minimizing total misclassification cost and a hybrid fitness function that uses an average value of the first two fitness functions to minimize total misclassification cost. Using simulated data sets representing three different distributions and four different misclassification cost matrices, we test the performance of the three fitness functions on a two-group classification problem. Our results indicate that the posterior probability-based misclassification cost minimizing function and the hybrid fitness function are less prone to training data over fitting, but direct misclassification cost minimizing fitness function provides the lowest overall misclassification cost in training tests. For holdout sample tests, when cost asymmetries are low (less than or equal to a ratio of 1:2), the hybrid misclassification cost minimizing fitness function yields the best results; however, when cost asymmetries are high (equal or greater than a ratio of 1:4), the total misclassification cost minimizing function provides the best results. We validate our findings using a real-world data on a bankruptcy prediction problem.Journal of the Operational Research Society (2009) 60, 1123-1134. doi:10.1057/palgrave.jors.2602641; published online 25 June 2008.

AB - We study three different approaches to formulate a misclassification cost minimizing genetic algorithm (GA) fitness function for a GA-neural network classifier. These three different approaches include a fitness function that directly minimizes total misclassification cost, a fitness function that uses posterior probability for minimizing total misclassification cost and a hybrid fitness function that uses an average value of the first two fitness functions to minimize total misclassification cost. Using simulated data sets representing three different distributions and four different misclassification cost matrices, we test the performance of the three fitness functions on a two-group classification problem. Our results indicate that the posterior probability-based misclassification cost minimizing function and the hybrid fitness function are less prone to training data over fitting, but direct misclassification cost minimizing fitness function provides the lowest overall misclassification cost in training tests. For holdout sample tests, when cost asymmetries are low (less than or equal to a ratio of 1:2), the hybrid misclassification cost minimizing fitness function yields the best results; however, when cost asymmetries are high (equal or greater than a ratio of 1:4), the total misclassification cost minimizing function provides the best results. We validate our findings using a real-world data on a bankruptcy prediction problem.Journal of the Operational Research Society (2009) 60, 1123-1134. doi:10.1057/palgrave.jors.2602641; published online 25 June 2008.

UR - http://www.scopus.com/inward/record.url?scp=68149150914&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=68149150914&partnerID=8YFLogxK

U2 - 10.1057/palgrave.jors.2602641

DO - 10.1057/palgrave.jors.2602641

M3 - Article

AN - SCOPUS:68149150914

VL - 60

SP - 1123

EP - 1134

JO - Journal of the Operational Research Society

JF - Journal of the Operational Research Society

SN - 0160-5682

IS - 8

ER -