Authors:
Roberto Ruiz
;
José C. Riquelme
and
Jesús S. Aguilar-Ruiz
Affiliation:
University of Seville, Spain
Keyword(s):
Data mining, preprocessing technics, database reduction, feature selection, data set editing.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Biomedical Engineering
;
Business Analytics
;
Data Engineering
;
Data Mining
;
Databases and Information Systems Integration
;
Datamining
;
Enterprise Information Systems
;
Health Information Systems
;
Sensor Networks
;
Signal Processing
;
Soft Computing
Abstract:
Progress in digital data acquisition and storage technology has resulted in the growth of huge databases. Nevertheless, these techniques often have high computational cost. Then, it is advisable to apply a preprocessing phase to reduce the time complexity. These preprocessing techniques are fundamentally oriented to either of the next goals: horizontal reduction of the databases or feature selection; and vertical reduction or editing. In this paper we present a new proposal to reduce databases applying sequentially vertical and horizontal reduction technics. They are based in our original works, and they use a projection concept as a method to choose examples and representative features. Results are very satisfactory, because the reduced database offers the same intrinsic performance for the later application of classification techniques with low computational resources.