Graph Builder Preferences: Middle 90% (or 99%)

[hogi]

☐ cool new feature
☑ could help many users!

☐ removes a „bug“

☐ nice to have

☐ nobody needs it

What inspired this wish list request? 

Some while ago, I found this post in the community:
In graph builder, is there a way to set the default color code gradient range to "Middle 90%"? 

by @tsg.

Up to now, there was no reply - and I could not find a related wish ...

What is the improvement you would like to see? 

Possibility to set the default range of a color gradient to Middle 90%.

Alternative options: 1% - 99%?

Why is this idea important?

The middle 90% option is amazingly useful - outliers don't destroy the color settings. Especially for exploratory data  analysis of a "dirty" dataset - e.g. by using the Column Switcher!

But at the moment, the user has to enable the setting manually - and concerning Column Switcher it's disabled after every single switch. Argh!

The possibility to set the default range of a color gradient to Middle 99% would solve the issue

more wishes submitted by  

[ehchandlerjr]

@Sarah-Sylvestre Not sure if people commenting as opposed to Kudos-ing helps this make it onto the list, but I second this, especially a wider range of values for the "middle x%". 

[hogi]

Thanks, @ehchandlerjr for supporting the wish.

A Graph Builder with a default color range option "1-99%" or +/- N-sigma will revolutionize the usability of the Graph Builder

[ehchandlerjr]

Yea 100%. There is a way to set your own levels, like a segmented gradient function. But it's (relatively) labor intensive and not dynamic. While I think that solution is also a necessity, your solution I think would be much more generally applicable. I have a plot where I have data points from 1:30 - 60:00 mis, and a couple of 44 he points. Having that middle x% where x is arbitrary would be . Hope it gets through! 

[hogi]

Here is a script which switches forth and back between "middle 90%" and "Default" simultaneously for all Gradients in the current report.

I will add it to Graph Builder Toolbar  ...

View more...

Names Default to Here(1);
If( !Namespace Exists( "hogi.GradientSettingsNS" ), New Namespace( "hogi.GradientSettingsNS" ) );

//toggle
Try( hogi.GradientSettingsNS:ninetyPercent,hogi.GradientSettingsNS:ninetyPercent = 0 );
hogi.GradientSettingsNS:ninetyPercent = Not( hogi.GradientSettingsNS:ninetyPercent );

Local({gbs, legendServers,myLegendItems, myGradients},

gbs=Filter Each({myItem},gb = Associative Array((current report() << xpath ("//OutlineBox")) << get scriptable object()) << get keys(),Not(isEmpty(myItem)));
legendServers = gbs << Get Legend Server();
myLegendItems = legendServers <<Get Legend Items();
myGradients= {};

For Each( {myGroup1}, myLegendItems,
	For Each( {myGroup2}, myGroup1, 
		For Each( {myItem}, myGroup2, If( myItem << get type == "Gradient", Insert Into( myGradients, myItem ) ) )
	)
);

for each({myItem}, myGradients, 
if(hogi.GradientSettingsNS:ninetyPercent,
myItem <<Set Properties({gradient( {Range Type( "Middle 90%" )} )} ),
myItem <<Set Properties({gradient( {Range Type( "Default" )} )} ));
))

[ehchandlerjr]

This is great! Thanks! 

[LoveDataViz7]

Similar older request was for 2.5% - 97.5% here:  Add histogram and sliders in gradient settings within graph builder, and change ... - JMP User Commu...

I like the histogram option suggested there as well, as it is often tough to know which manual limits to add and I end up opening a separate window for the distribution as a reference.  Anything that defaults to a better pre-set "most of the data" visual would be very helpful.  10-90% would work great as well.

[hogi]

so, time is ticking since 2020 ...

[hogi]

[edit / striketrough text: I "forgot" that the wish is for the GRADIENT - not for an axis range.
So, no data points will be hidden
@ehchandlerjr, thanks for the hint]

I don't want to upset anybody who voted ...
(There are just a few wishes in the Wish list with more Kudos *)

Nevertheless, as for me, the wish could be closed.

most probably I will not use a middle 90% setting in the preferences:

actually, it's a good idea to always start with a plot that shows ALL datapoint - nothing hidden.

*)

[ehchandlerjr]

@hogi to clarify though, this is not removing data, but simply making sure it doesn't mess up the gradient, correct? In my mind this is as much about quickly conveying information at a meeting as it is for the users own analysis, if not more.