A bit unconventional, but I have a need to build a heatmap with 5 variables. Basically you can think of this as being a typical heatmap with three (bottom, left, and Z) but then replicated in columns and rows across some other variable. I have accomplished something similar with subplots and heatmaps but it gets expensive to redraw when I have 20 or 30 (or more) of these. Is there any way to build a single heatmap showing four or five dimensions? If I could define labels for both the bottom and top and left and right all on the same plot I could build the data structure but I can’t figure that out.
Thanks!
How would you fit more than four dimensions into a 3D plot?
I’m currently building 4D plots (XYZ+Value), and the only way to visualize more data is to build more subplots (or separate plots) with the additional traces.
I’d like to see how you approach this.
I would use position for three of the dimensions, color for the fourth, and then size for the fifth. But that gets especially hard to visualize in 3D space.
In my opinion, higher-dimensionality data is more easily interpreted with a heatmap for the type of analysis I am doing.
We do exactly that, plot several 2D plots alongside.
It gets heavy though, because they have lots of points each (computed on the browser).
I am unable to provide an example right now, but if anyone ever needs this capability you can actually add three traces to the heatmap - one covering all your data across all variables as the Z, then add additional traces for the “top” and “right” labels with the ticks set manually.
1 Like