Visualizing the effects of scale and geography in multivariate comparison

Our research investigates the sensitivities and complexities of visualizing multivariate data over multiple scales with the consideration of local geography. We investigate this in the context of creating geodemographic classifications, where multivariate comparison for the variable selection process is an important, yet time-consuming and intensive process. We propose a visual interactive approach which allows skewed variables and those with strong correlations to be quickly identified and investigated and the geography of multi-scale correlation to be explored. Our objective is to present comprehensive documentation of the parameter space prior to the development of the visualization tools to help explore it.