Correlation Graphics and Cluster Maps
Marco Cremonini · 2024
Correlation graphics are a family of graphics aimed at showing the possible statistical correlation between variables. From the statistical correlation index is then possible to analyze the possible cause–effect relationship between two variables. This chapter deals with a graphic type that goes often under the name of cluster map and represents an extension of traditional heatmaps, enhancing them with graphical elements derived from clusterization methods, which are statistical methods aimed at grouping observations based on similarity or correlation metrics. Before introducing correlation matrixes for Python, and assuming as given the basic knowledge about them, it should be clarified that what readers will compute are so-called Pearson correlation matrixes, which aim at measuring the linear correlation degree between continuous variables having normal distribution. Seaborn offers another interesting visualization for correlation matrixes, again as a smart variant of a traditional graphic.