Prediction of Social Events Using a Classifier System
Hilty, John A.
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/72115
Description
Title
Prediction of Social Events Using a Classifier System
Author(s)
Hilty, John A.
Issue Date
1992
Doctoral Committee Chair(s)
Carnevale, P.,
Department of Study
Psychology
Discipline
Psychology
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
Psychology, Social
Artificial Intelligence
Computer Science
Abstract
This research investigated the potential of a technique from artificial intelligence, the classifier system, as a tool for time-series analyses in the social sciences. This technique is intriguing because of its similarity to some processes of evolution and the optimization of living organisms; there is increased interest in such models because of their robust self-organization and capacity to escape from "brittleness" (Holland, 1986). To explore these properties, a classifier system was designed and applied to four archival databases to predict the following events: conflict level between the United States and the U.S.S.R., emotions of a diarist, voting behavior in the 1960 presidential election, and outcomes of disputes that were subject to professional mediation. This classifier system is similar to others that have been used in the past to predict events (Schrodt, 1986; Smith and Green, 1987; Riolo, 1988; de la Maza, 1989), but it possesses several innovative characteristics: both Darwinian and non-Darwinian algorithms of rule generation, collective decision criteria, strength types that derive from the performance dimensions of frequency, accuracy, uniqueness, activity, and base rate. Furthermore, several aggregation techniques are proposed to resolve conflicts across replications, and several similarity coefficients and distance metrics are proposed to evaluate various aspects of predictive performance. The performance of the best model of the classifier system was compared with several time-series techniques from statistics: multiple regression, probit, discriminant analysis, Markov analysis, ARIMA, and others. It was found that the predictive performance of the classifier system was slightly to moderately greater than the other time-series techniques for most measures for three of four archival databases. Supplementary analyses provided strong support for the innovative characteristics of the classifier system, assuming that they are combined appropriately. Characteristics of causality and philosophy of science issues, as they relate to this research, were explored.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
