Classification of Interstate Conflict Outcomes Using a Bootstrapped CLS Algorithm
Loading...
Date
1987
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
"The CLS aglorithm is an inductive technique developed in artificial intelligence for generating classification trees from a set of data. These trees are similar to those used in expert systems; the advantage of the CLS algorithm is that the trees are being generated automatically rather than via human experts. This paper applies a bootstrapped version of CLS to the Butterworth Interstate Conflict data set. By generating a number of classification trees from randomly selected subsets of the complete data set, the variables which are the most effective in correctly classifying the cases can be identified and the degree of unpredictability in the data can be ascertained by computing the accuracy of the tree in classifying those cases not in the training set. In general, the technique works very well; the original set of 38 independent variables can be reduced to five or fewer with almost no loss of classification accuracy. Classification trees generated with these variable have 95%-100% accuracy when fitted into the entire set, and average 50%-60% accuracy when tested against validation samples in split-sample tests. Unlike existing statistical techniques, the knowledge representation structures inductively constructed by bootstrapped CLS are plausible models of human inductive theorizing since they fit within the known cognitive constraints of the brain."
Description
Keywords
artificial intelligence, common pool resources, conflict--models