Figure 1

BiFold as a joint visualization. (Middle panel) The classic Southern Women Dataset [26], where Davis has arranged the table to highlight the social group structures. (Left panel) Independent MDS representations can be created for the women (using inter-row distance) or for the events (using inter-column distance), but those plots cannot be directly merged because the coordinate axes are not same. (Right panel) The BiFold representation gives coordinate representation against a common basis. Note that the significant “clusters” observable in the sorted data matrix (colored rectangles) are matched to spatially clustered sets (colored ellipses) in the BiFold plot.