collapse layer selector in mode "view" & scale trans functions (pseudo_log with sigma)

[xtimbeau]

Hi, thanks for the package and the new v4 option, love the scale stuff (but you should add trans for neg/pos range like pseudo log).

Is there a way to collapse the layer selector in interactive mode (tmap-mode("view"). It stays open and occupy a lot of space on a standard map.

[mtennekes]

Hi @xtimbeau thanks for letting me know. This is a view-only option which we still have to migrate. Not a lot of work though :-)

Besides that, I am curious about your suggestion related to the scale. Can you give me an example?

[xtimbeau]

hi @mtennekes, you can have a look at the scales package (https://scales.r-lib.org/reference/log_trans.html). Such transformation is very useful for color scales when the range of colors is negative to positive but when a linear scale gives too much room to high absolute value. Using that transform (and the cut parameter associated) allows to improve contrast.

[mtennekes]

Ok, I borrowed some code from scales::pseudo_log_trans, and implemented a new trans, called "pseudo_log".

Please check if it makes sense @xtimbeau :

World$HPI2 = World$HPI - 30
tm_shape(World) +
	tm_polygons("HPI2", fill.scale = tm_scale_continuous())
tm_shape(World) +
	tm_polygons("HPI2", fill.scale = tm_scale_continuous(trans = "pseudo_log"))

The functions from scales has two parameters, sigma and base. Are both useful?

Could you provide a data example for this scale would be useful?

[xtimbeau]

@mtennekes great ! I'll test it and try to factor a nice example.

[xtimbeau]

@mtennekes sigma is quite useful as it sets the cut for when the log is smoothed to a linear function. Pseudo log looks like a log above a few signa (or below a few -sigma) but is linear between -sigma and sigma. Base is also relevant because of that non log linearity.

[mtennekes]

Collapse has already been set to TRUE by default (forgot to mention earlier).

I've implemented tm_scale_continuous_pseudo_log, with sigma and base. Indeed very handy @xtimbeau this sigma parameter!

Also added tm_scale_continuous_log10, tm_scale_continuous_log2, and tm_scale_continuous_log with base. Please check. The only difference between log, log2, log10, is that with a base other than exp(1), the labels will be formatted as base^x.