Degree Type
Thesis
Date of Award
2011
Degree Name
Master of Science
Department
Industrial and Manufacturing Systems Engineering
First Advisor
Sigurdur Olafsson
Abstract
Decision trees are a useful tool to help in the extraction of information from databases, but all too often this ability is clouded by the complexity of the tree structure resulting from the decision tree algorithm. Methods such as tree pruning, attribute selection, and most recently, instance selection, currently exist to simplify the decision tree structure. We present an alternative instance selection procedure for simplifying decision trees that improves upon previous methods by increasing the quality of the space to be traversed for finding an acceptably simplified decision tree through the identification and grouping of important instances. Experimental results from this procedure are then presented and compared to decision trees with no prior simplification effort applied. We show that in some cases we are indeed able to identify important group of instances, and subsequently are able to generate a high quality solution space for finding simplified decision trees.
DOI
https://doi.org/10.31274/etd-180810-1522
Copyright Owner
Walter Dean Bennette
Copyright Date
2011
Language
en
Date Available
2012-04-30
File Format
application/pdf
File Size
70 pages
Recommended Citation
Bennette, Walter Dean, "Instance selection for simplified decision trees through the generation and selection of instance candidate subsets" (2011). Graduate Theses and Dissertations. 12084.
https://lib.dr.iastate.edu/etd/12084