Cardiff University | Prifysgol Caerdydd ORCA
Online Research @ Cardiff 
WelshClear Cookie - decide language by browser settings

Deriving prediction rules from co-occurrence data: A method based on Boolean algebra

Von Hecker, Ulrich 2001. Deriving prediction rules from co-occurrence data: A method based on Boolean algebra. Psychologische Beiträge 43 (2) , pp. 293-311.

Full text not available from this repository.

Abstract

This paper deals with a Boolean method of deriving prediction rules from co-occurrence data. This method makes use of a Boolean minimization algorithm, applying a so-called eliminative strategy. The rationale is to treat the entire data matrix as a single, complex Boolean term which is then simplified, terminating with its minimal form. The resulting term can easily be interpreted as a sequence of subterms, each of which is equivalent to one particular prediction rule of the kind as defined within the FPA framework. It is discussed in which way such minimization methods can be applied within the FPA framework. Statistical problems are also considered, e.g., the evaluation of the predictive power of single rules in terms of PRE measures. The procedures are illustrated by an example from clinical neuropsychology. A binary data set about presence vs. absence of spontaneous speech symptoms in a number of brain-injured patients is examined in order to predict the main aphasic syndromes by specific patterns of speech impairment. It is demonstrated that the proposed set of Boolean methods leads to some rules with high diagnostic relevance for the prediction of the Wernicke and Broca types of aphasia.

Item Type: Article
Date Type: Publication
Status: Published
Schools: Psychology
Subjects: B Philosophy. Psychology. Religion > BC Logic
B Philosophy. Psychology. Religion > BF Psychology
Uncontrolled Keywords: Boolean algebra, minimization procedures, prediction of aphasic syndromes
Publisher: Pabst Science Publishers
ISSN: 0033-3018
Last Modified: 04 Jun 2017 04:09
URI: http://orca.cf.ac.uk/id/eprint/33603

Actions (repository staff only)

Edit Item Edit Item