topic Re: Group multiple column in Discussions

Group multiple column

ConcurrentStudy — Sat, 10 Jun 2023 23:58:25 GMT

Hi,

I'm novice in JMP. Starting to learn JMP. I'm trying to group a column and count the number of row.

Below is the example to clarify my question. Basically i have 4 unit tested with different pattern. Each pattern will output failure.

How can i group them based on pattern. Any help will be appreciated. Thanks

Example raw data:

	TEST A	TEST A FAIL OUTPUT	TEST B	TEST B FAIL OUTPUT	TEST C	TEST C FAIL OUTPUT
UNIT A	A0 PATTERN	A1
UNIT B	A0 PATTERN	A2	B0 PATTERN	B1	C0 PATTERN	C1
UNIT C	A0 PATTERN	A1	B1 PATTERN	B2
UNIT D			B1 PATTERN	B3	C0 PATTERN	C2

Output needed:

A0 PATTERN	A1	2
	A2	1
B0 PATTERN	B1	1
B1 PATTERN	B2	1
	B3	1
C0 PATTERN	C1	1
	C2	1

Re: Group multiple column

txnelson — Thu, 15 Dec 2022 14:16:15 GMT

The solution is to stack the data and then to use the Tabulate Platform to produce the required output.

Starting with your data table

Go to Tables=>Stack

Fill out the Dialog Box as shown

Click on OK and it will give you the table as previewed

Now go to Analyze=>Tabulate

Drag the Data column to the Drop Zone for Rows

Drag the Data 2 to the far edge of the above dragged Data column and you will have the report you want

Re: Group multiple column

GregF_JMP — Thu, 15 Dec 2022 14:51:44 GMT

Hello, welcome to JMP.

The table operations are very powerful and will help in this instance. The icons on the Table menu items help describe the actions that reshape data.

There are a number of ways to accomplish your objective, but here is one.

1) Your example data should have column headers before moving to JMP, assume First column is called "Part" in upper left cell
"Copy", "Paste with Column headers" into a blank JMP data table

2) Execute the Table Operation "Stack" (New in JMP 17, the preview window helps with real time feedback of potential output)

Select all Columns except "Part", press "stack columns" button to select six columns with test information to be stacked
Check "multiple series stack", set "number of series" to 2
De-select "Contiguous"
Check "Eliminate Missing Rows" (optional- it's easy to delete the non-failing test items later)
Optionally, for non-stacked columns you can "keep all" in case Part/Unit ID will be helpful for other analysis.
Press Ok

3) On the table generated by the "Stack" operation, execute the Table Operation "Summary"

Select Columns Data and Data 2, Press "Group" button so both columns are listed in the group box. This will group all combinations of these values. (control-click to select multiple items or do in two steps)
Uncheck "Link to original data table"
Press OK

Your requested output

Re: Group multiple column

ConcurrentStudy — Fri, 16 Dec 2022 02:09:56 GMT

Thanks

I have tried it works on 3 or 4 pattern.

But when i include all my pattern column around 40++ with it output also around 40++. it will pop up this error.

I tried to make all my column to ordinal or nominal but not able to solve this one.

Any idea why it produce such error?

Re: Group multiple column

txnelson — Fri, 16 Dec 2022 06:09:21 GMT

If you change all of the columns to character, the stack will then work.