Home // DBKDA 2015, The Seventh International Conference on Advances in Databases, Knowledge, and Data Applications // View article


On the Number of Rules and Conditions in Mining Incomoplete Data with Lost Values and Attribute-concept Values

Authors:
Jerzy W. Grzymala-Busse
Patrick G. Clark

Keywords: Data mining; rough set theory; probabilistic approximations; MLEM2 rule induction algorithm; lost values; attribute-concept values

Abstract:
In mining incomplete data, we have a choice for interpretation of missing attribute values. In this paper, we consider two such interpretations: lost values and attribute-concept values. To measure the number of conditions and rules for each interpretation, we conducted experiments on eight incomplete data sets using three kinds of probabilistic approximations: singleton, subset and concept, with eleven values of probability. Using a 5% significance level, the results show that the number of rules is always smaller for attribute-concept values than for lost values. Additionally, the total number of conditions is smaller for attribute-concept values than for lost values for seven out of eight data sets.

Pages: 121 to 126

Copyright: Copyright (c) IARIA, 2015

Publication date: May 24, 2015

Published in: conference

ISSN: 2308-4332

ISBN: 978-1-61208-408-4

Location: Rome, Italy

Dates: from May 24, 2015 to May 29, 2015