Abstract:

In this paper we propose a method for decision making based on performance evaluation. We considered the case of the 28 E.U. member states, for which macroeconomic indicators of economic growth, current account balance, labour productivity, foreign direct investments and general government debt were used. The E.U. economic performance was evaluated by applying a Hierarchical Cluster and the results suggested that based on labour productivity, current account balance and GDP growth rate, the 28 E.U. member states can be classified into two main clusters, one corresponding to high economic performant countries and the second one to countries with lower economic performances. Based on this classification, we then built two CHAID classification trees and tested their prediction ability to correctly classify the 28 E.U. countries into high and lower economic performances countries. Our results suggested that when replacing the macroeconomic indicators of the prediction model with the first principal component of the initial data matrix, the prediction performance of the new CHAID model improved and reached an accuracy of prediction of 89%.