How to group points in a plot and assign categories in data table (3 or more categories)

[#1]

How can one categorize a visual data set into user defined categories in the original data table?

To explain this further. Say I have a scatter plot, and I want to select certain group of points, and call it Category 1, then somewhere else, select another group, call it category 2, and so on. The goal is to make an interval plots of these 'categories' and see how the means etc vary.

Currently, I can use Data View for selection, and fill, say, Tag 1, then repeat with new selection, and call it Tag 2, and so on. Very manual.

In an ideal case, I should be able to "tag" group of points on the plot, choose a destination column in the data table, and perhaps have color option to keep each of the groups visually distinct.

Any pointer on how to make this easy would be much appreciated!

Thanks and Many Kudos :)

[#2 · Accepted solution]

1. Select the data points on the graph you want to categorize

2. Right click on one of the selected points

3. Select "Name Selection in Column"

4. Complete the dialog box with the column name and value to use

5. Click OK

[#3 · Accepted solution]

Hi @Kapoor,

Building on what Jim suggested, when you use "Name Selection in Column" be sure you use the same name for the variable each time you make a selection, and also remove the value for the "Unselected" rows (that is, delete the 0). This way you can build your new column with just the two "Name Selection in Column" commands.

@julian

[#4]

Thanks Jim and Julian. Your combined solution works well !

As an FYI, in JMP 13,

After right click on data point, I need to go "Rows" before getting to "Name Selection in Column"

There was a second question: How to color the selection so one can demarcate the selected clusters.

The answer is same - from "Rows" choose "Rows Colors"

(One caveat though, that would not work if  "Color" data zone has been triggered)

Thanks to all again!

[#5]

Hi @Kapoor,

That's correct, if you're doing your selections in Graph Builder you need to go to the Rows submenu to find Name Selection in Column. In most other platforms it's behind just the Right Click. 

Oops, I did miss that second question. Yes, using Rows > Row colors is one easy way to color the points. Alternatively, once you've named all the selections you could go to the Rows menu > Color or Mark by Column to set the Row states all at once (and here you could also set different markers for the rows). This might be helpful if you are naming many selections and don't want to take the time to color each one separately.  You're right that the Graph Builder Color role will override these colors, which is why having different markers might be helpful. 

@julian

[#6]

Does anyone has JMP script for selecting group of points i.e. rows and categorizing with different name? Also, if the same group comes again in the same dataset increment the category name by 1. Any help is really appreciate.

Example below:

Row	Output	Category
1	0.5	OFF1
2	0.5	OFF1
3	1.2	0
4	1.5	0
5	10	ON1
6	10	ON1
7	7.5	0
8	6	0
9	0.5	OFF2
10	0.5	OFF2

[#7]

How are you defining your "Group of Points"?

[#8]

Currently I am selecting manually by looking at the data. But below is what I want.

Starting to scan from Row 1 and defining the criteria like output <1. as soon as it sees it. Category1. Then the creteria changes because the output changes and based on output change the creteria changes i.e.9< output <11. and so on.

[#9]

Selecting points in a graph, has 2 dimensions. So when you talk about "Row", are you talking about the minimum value on the X axis? And then you look for when the value of the Y axis when you move up higher on the X Axis? So when the Y axis value is found, then those points are declared a Category(Grouping/Bin). Then the definition of category changes to a new group definition and moves on.....is this what you are saying?

[#10]

yes exactly.