Spatial Data Mining: From Theory to Practice with Free Software

Abstract

One of the main challenges in spatial data mining is to automate the data preparation tasks, which consume more than 60% of the effort and time required for knowledge discovery in geographic databases. In this paper we present an extension of the classical open source data mining toolkit WEKA to support automatic geographic data preprocessing. We propose Weka- GDPM+, which is interoperable with all geographic information systems constructed under OGC specifications. We tested Weka-GDPM+ with geographic databases stored into PostGIS, which is an open source GDBMS and is implemented according to OGC standards.