Validating data warehouse quality metrics using PCA

Abstract

During the last decades researches, Data Warehouse is mainly concentrated on Quality. The best approach to quality evaluation goes through determining effective metrics on schemas. However, this set of metrics contains some redundant metrics. Generally, PCA (Principal Component Analysis) is used for defining principal metrics in the domain. In this study, we used PCA for dimensionality reduction on a set of 41 schemas, and we find out that instead of seven metrics [3], only 3 metrics were extracted as principal components. Our empirical validation experiment showed us that three principal components out of seven proposed metrics [3] seem to be practical indicators of the Quality Model for Data Warehouses. © 2012 Springer-Verlag.