Manual for EAR4 and CAAR Weka Plugins, Case-Based Regression and Ensembles of Adaptations, Version 1

Abstract

EAR4 and CAAR are lazy learners applying the case-based reasoning (CBR) paradigm to numerical prediction tasks. Both augment standard instance-based learning methods by applying automatically generated case adaptation rules to adjust solutions of prior cases, and both apply ensembles of the generated rules. CAAR augments the EAR approach with a richer treatment of case context, more context-aware rule generation, and context-sensitive ranking of the generated adaptation rules. This manual describes installation and use of plugins enabling use of EAR4 and CAAR within the Weka workbench for machine learning.