783786_en-US

liliconcom

Why is this DropZone called "Color"

Day 1 with JMP and a user wants to generate a point plot with 2 different colors for male and female:

 

The easy solution: use the Color Dropzone

The surprise: The Overlay does the same.

The second surprise: "Color" can do more than just "color":
For every entry in the legend, the user gets the possibility to define the Marker Style as well - as well as Size and Transparency.

A lazy user doesn't configure the colors manually - again and again for every single plot and subplot.

 

One alternative - definitely not the best!:  row states to specify Colors and Markers.

disadvantages: 
- the legend in Graph Builder doesn't fit
- row states are static. When the user changes the values in the table, they have to correct the row states manually. Otherwise the plots will be wrong!

Another alternative - by far better:

define suitable colors via Column Property: Value Columns
Pros:
- meaningful legends - automatically (!)

- separate settings for every column (not via rows)

- I can control (via the Overlay and Color dropzone) where I want to use the setting - and for wich plots I DON'T want to use the setting : )

- I can overwrite the setting for individual entries via the legend right click menu

The 3rd surprise: there is no Column Property: Value Markers *)

 

Day 2 - the JMP user detects the subtle (and not so subtle) differences between the  Color and Overlay Dropzones:
by adding a smoother:

Now, let's add Color to the plot
suprise 4: with Color via Color, there is just a single fit curve:

 

... this is when the user remembers surprise #1, tries Overlay, and falls in love with the Overlay dropzone:

 

So, for scatter plots there is no apparent difference between Color and Overlay - but for other plots it is !!!
And often Overlay is THE way to add color.
Another example:

vs.

 

Now let's generate another points plot - where there is a difference between Color and Overlay:
Color just adds color to the points, Overlay generates separate plots for every color, overlays them

- and by doing so: hides some girls. -> dangerous!

vs. 

 

Let's keep the plot as a stacked bubble plot, and with the findings from surprise #2, let's now define Markers for different values
... with the help of the Color Dropzone

Data Exploration and VisualizationRe: Why is this DropZone called "Color"

Day 3: The user likes the non.overlapping points, wants to add some Color (via age)  .. and get the legend right.

 

 

Re: Why is this DropZone called "Color"

differences between Overlay and Color:

:folded_hands: Graph Builder: support more than 2 'Color' columns
- > available since JMP19

 

Graph Builder - Overlay by multiple columns :
not yet available, still collecting Kudos ...

Re: Why is this DropZone called "Color"

new Column Property: Value Markers *)
->  again open for voting

Just imagine how helpful it will be:
- meaningful legends in Graph Builder (compared to the alternative via row states)

- fully automatic + consistent  (compared to assigning markers manually via the legend)

interested?
Follow the link and support the wish with a Kudo.

Re: Why is this DropZone called "Color"

for "no zig-zag lines", there exists a wish to automatically implement the functionality in Graph Builder:
Graph Builder: Line - possibility to add breaks 

Re: Why is this DropZone called "Color"

I need this script...please share!

Re: Why is this DropZone called "Color"

Did you know that Overlay Encoding can be used to Overrule the Color Dropzone:
[age used for the Color Dropzone - but the Overlay encoding is set to Color
- so, color of the points is determined by the Overlay settings, NOT by the Color Dropzone.
changing the color settings for age doesn't have an effect on the colors of the Graph]

 

 Names Default to Here(1);
dt = Open( "$SAMPLE_DATA/Big Class.jmp" );
Graph Builder(
	Variables(
		X( :height ),
		Y( :weight ),
		Overlay( :sex, Overlay Encoding( "Color" ) ),
		Color( :age )
	),
	Elements( Points( X, Y, Legend( 1 ) ) ),
	SendToReport(
		Dispatch( {}, "400", ScaleBox,
			{Legend Model( 1, Properties( 6, {Marker( "Triangle" )}, Item ID( "female", 1 ) ) )}
		)
	)
);

Re: Why is this DropZone called "Color"

This rule holds ...  till you use BLACK as Overlay color:
IF Overlay color is black, JMP will use the color from the Color Dropzone - despite Overlay Encoding = Color.

 

 

 

Names Default to Here(1);
dt = Open( "$SAMPLE_DATA/Big Class.jmp" );
Graph Builder(
	Variables(
		X( :height ),
		Y( :weight ),
		Overlay( :sex, Overlay Encoding( "Color" ) ),
		Color( :age )
	),
	Elements( Points( X, Y, Legend( 1 ) ) ),
	SendToReport(
		Dispatch( {}, "400", ScaleBox,
			{Legend Model(
				1,
				Properties(
					1,
					{Line Color( 0 ), Marker( "FilledCircle" ), Marker Size( 6 )},
					Item ID( "F", 1 )
				)
			)}
		)
	)
);

 

Re: Why is this DropZone called "Color"

Another case where the above rule doesn't hold: Smoother, Bar, Contour , Box Plot  & Line graphs *)

 

For Smoother and Line Graphs, the COLOR Dropzone controls the Color, independent of the Overlay Encoding for the Overlay.

This fact can lead to plots with inconsistent color settings:

 

*) Please note:

Line of Fit and Ellipse  behave like points [but without the possibility to switch to Color by setting the Overlay color to black] 

Area Graph seems to be a combination of both worlds:

 

Re: Why is this DropZone called "Color"

in Pictures from the Gallery#9, @scwise mentions this mismatch between Overlay encoding @ 21 min .

Re: Why is this DropZone called "Color"

I really like JMP's approach, where a user defines some default settings via column properties - and every new graph is perfect from the start - nothing to change manually. Unfortunately, this approach doesn't work up to 100%.
In most cases I have to manually change the settings to get the graph I want.

 

Manually changing the settings by right-clicking on the legend is very time consuming. This video demonstrates an alternative approach: Basically ignore the default settings and give the user an interactive GUI to precisely specify the settings for each individual element. Compared to right-clicking on the legend items, it can be orders of magnitude faster to get the graph you want.
The drawback:
- too much flexibility - which can lead to inconsistent settings (can be intentional, like in the video - may be unintentional)
- I have not yet managed to create a meaningful legend for the final graph.

 

Re: Why is this DropZone called "Color"

I think of "color" as being strictly color, but "overlay" as being the equivalent of the old transparencies-on-a-projector apparatus where you could "overlay" different plots on top of each other.  Yes, it's a little weird that overlay also drives the color, but I'm glad it does.  Overlay is also one of my favorite features of Graph Builder because it lets me break line plots wherever I want using a specific variable.

Re: Why is this DropZone called "Color"

---

@hogi wrote:

Overlay takes away most of of Color's superpowers - just Color and Transparency are left:

 

---

oh yes - and actually, there are some cases where the Color dropzone takes the full control for Transparency.
There is a Transparency setting for the individual Overlay entries, but without effect:
Transparency for Overlay Group? 

 

I hope in a future version of JMP the user will get more flexibility to adjust which settings are controlled by Color or Overlay .

Re: Why is this DropZone called "Color"

Color is much more than just "Color", see :
Why is this DropZone called "Color" 

---

@hogi wrote:

Day 1 ...

The second surprise: "Color" can do more than just "color":
For every entry in the legend, the user gets the possibility to define the Marker Style as well - as well as Size and Transparency.

zoom out view:

 

For Color together with Overlay, you are right.
Overlay takes away most of of Color's superpowers - just Color and Transparency are left:

Re: Why is this DropZone called "Color"

Interesting:

Overlay Dropzone: 
The user can adjust the Line Color and the Fill Color for every item:

Color Dropzone:
- it says Fill color, but Graph Builder changes the Line Color.

What is the purpose of Fill Pattern?

no effect (A)  - there is not even a fill pattern:

*) no effect (B)  - after applying a Fill Color via the "box plot" settings:

[please note the difference between the legend and the plot]

 

Issue with Overlay    and   Color by Theme:
The color is applied as Line Color and as Fill Color:

Is there a trick to adjust the line color only?

Re: Why is this DropZone called "Color"

Graph Builder: support more than 2 'Color' columns (:green_heart:) 
-> now available in JMP19   : )

Re: Why is this DropZone called "Color"

How to synchronize color settings in JMP?
(same color for points and lines | same color for the left and right plot ...)

 

by far the best option:  Color Properties / Value Colors & Color Gradient: 
very easy to define -> all plots share the same settings | no "shifts" caused by subsets and local data filters

 

Within a plot window, one can specify a color setting which is applied to all plot types (lines, points) and subplots.
via red triangle/Color Settings:

It is important to note the following when the user wants to apply color settings manually:

1. If only one entry is visible in the legend ("M" is displayed as points and lines, but there is only one item in the legend),
   the setting is applied to every corresponding plot (points, lines and other regions of the plot) -> synchronized changes *)
2. If there are more entries visible in the legend (e.g. two 'M's), the changed setting will only apply to one plot type/subplot.
3. The transition from Mode 1 to Mode 2 is irreversible. Once a setting has been changed, it is not possible to switch back to mode 1. Not even Revert is enough to reset the setting (useful workaround: open the script, delete the settings and run the script)
   
   

    

4. revert:
   

    

*) this is not a bug, it is a feature

Re: Why is this DropZone called "Color"

related topic: assign a color to axis labels (to match the color used in the plot)

Until now, assigning a color to axis ticks and labels was complicated.
Users had to apply the color via the 'Properties' menu on the right.

 

This type of assignment is fragile: The setting is applied to the Nth axis Box.
Changing the plot layout ruins the assignment:

[after adding the second x axis, the labels on the right are black , labels on the second x axis are red]

 

With JMP19, there is now an easier and more robust way: just click 'Parallel Y Labels' in the red triangle menu.
This new functionality has been implemented Plotting several trends from sensor data that have different units (or scale) 
nice: the color will match the color in the plot