In my posts so far in trying to find more readable ways of presenting multiple series plots, I've tried multi-colors, and varying spectral maps. With multi-colors the lines may be easier to distinguish but harder to follow.
Eli, in a comment in the first post, suggested making curves respond to the pointer rolling over. This needs Java programming, which I can't do. But it occurred to me that an animated gif would achieve something of the same effect.
If each curve becomes, for a period, a continuous black line, then its path can be traced easily. Between times, the dots will separate out the local features. I'll switch to this more topical example - JAXA Ice extent:
Nice! That works for me!
ReplyDeleteBut actually the stripes are probably redundant now. It'd be better if the lines went back to being single colours, as long as those colours are well distinguished from black.
But then you run up against the colourblind thing again - to me red lines look very dark. Setting a lower bound of #40 for all of r/g/b would probably be fine though (so red would be #ff4040).
Kevin C
Nice.
ReplyDeleteRghcnV3 is up on cran.
windows binary and mac binary should follow in a few days.
Modest start, more to come.
Kevin C
ReplyDeleteThanks again for the feedback. I switched to a rainbow style as you suggested, and to Arctic Ice (TSI was getting boring). Hope that's better.
OK, that's awesome. It's a graph which I've seen many times, and for the first time can actually interpret.
ReplyDeleteEven the colours are better that normal - where the lines are well separated I can pick out all but 2005/06 and 2009/10/11. Lightening the colours has helped.
The animation solves the rest of the ambiguity.
Kevin C
Excellent
ReplyDeleteEli,
ReplyDeleteTheFordPrefect has shown me how to implement your Java rollover suggestion. That will be the next post.