I was thinking about using heatpams for some p-values and "freq-like" tables and I had some philosophical questions what would be the best way:
1) Should the colormap be discrete or continuous?
For example, 0, 0.1, 1 = green, also green, red (continuous)
Or should it be 0, 0.1, 1 = green, yellow, red (discrete)
I don't think many people would need that.
However, suggestion No.2 is more important
2) Could the colormap be scaled?
a) if it's scaled from MIN to MAX for every given table, it's providing "the best contrast", but two tables will not be comparable by looking at the colors alone
b) with the custom scaling, let's say from 0 to 1, all the tables will be comparable.
0.5 would be "yellow", both in the table that goes form 0 to 1
and in the table that goes from 0.4 to 1
I was thinking about using heatpams for some p-values and "freq-like" tables and I had some philosophical questions what would be the best way:
1) Should the colormap be discrete or continuous?
For example, 0, 0.1, 1 = green, also green, red (continuous) Or should it be 0, 0.1, 1 = green, yellow, red (discrete)
I don't think many people would need that.
However, suggestion No.2 is more important
2) Could the colormap be scaled?
a) if it's scaled from MIN to MAX for every given table, it's providing "the best contrast", but two tables will not be comparable by looking at the colors alone
b) with the custom scaling, let's say from 0 to 1, all the tables will be comparable.
0.5 would be "yellow", both in the table that goes form 0 to 1 and in the table that goes from 0.4 to 1
I think this would be useful
"Only label significant" = brilliant, I love it!