Making Subsets Smarter - Help Needed

[#1]

Hi Hive Mind,

I quite often make a subset of my data table.

My current method is

1. Find a value that is representative of my subset
2. Right click -> Select matching cells
3. Tables -> Subset à selected rows, all columns
4. Save file

Alternatively, I will

1. Use “Save as” to create a new file
2. Find a value that is representative of my subset
3. Right click -> Select matching cells
4. Invert selection
5. Delete Rows

I do this many times on a weekly basis… and it is painful…

So here is my idea that I need help with:

I want to create a script that when I have selected my matching cells I can click a button that will make a subset of the selected rows:

Here is my best shot so far (it doesn’t work)

Dt = current data set();

My_list = dt << Get Selected();

Dt << Subset(

                Selected Rows( 0 ),

                List;

                Selected columns only( 0 )

);

Do you have other ideas, or should I do something completely different?

Cheers,

Christian

[#2 · Accepted solution]

I have created "add-in" (it has many other shortcuts that's why it is an add-in) which basically just sets shortcut to run simple script like

Current Data Table() << Subset(Selected Rows(1), Selected Columns Only(0));

You could also do this by creating a toolbar and adding shortcut to it.

[#3]

Thanks for the answer Jarmo! I am creating an add-in with similar scripts that I find myself repeating! 

dot to comma is one of them XD

Names Default To Here( 1 );

dt = Current Data Table();

nr = nrows(dt);
col_names = dt << Get Selected Columns (String);// get column names (String);
for (i = 1, i <= nitems(col_names), i++,
    one_col = column(dt, col_names[i]);
    for (k = 1, k <= nr, k++,
        one_col[k] = substitute(one_col[k], ".", ",");    // Replace decimal with comma
        //one_col[k] = substitute(one_col[k], ",", ".");    // Replace comma with decimal
    );
    one_col << Data Type(Numeric);
    one_col << Set Modeling Type ("Continuous");
);

I got this from another community script.

[#4 · Accepted solution]

Your specified code only needs to be modified slightly to get what you want.

Dt = current data set();

Needs to be

Dt = current data table();

and the Subset Platform already can find all selected rows, and use that for the selection

Dt << Subset(
	Selected Rows( 1 ), 
	Selected columns( 0 )
);

Put it all together it looks like

Names Default To Here( 1 );
Dt = current data table();

Dt << Subset(
	Selected Rows( 1 ), 
	Selected columns( 0 )
);

As Jarmo suggests, you could turn this into a toolbar short cut.

[#5]

Thanks Jim, that was a very helpful answer. 

Enjoy your weekend!

[#6]

The given solution is already being used! thanks @txnelson and @jthi 

[#7]

Before creating an Add-In on your own, you might have a look at QuickSelect Toolbar 

The option which you describe is this one:

There are many more Shortcut included -  like Tagging (~ Name Selection in Column on Steroids)

... and there is also a Shortcut for Invert selection :)

[#8]

Actually, I started the Toolbar with a version like this one, returning a subset with just the selected rows and columns.

The current version includes now all columns - and is a linked subset. My work in Jmp would be much less efficient without it ...

application 1) select some points in a graph -> generate the subset

application 2) select some rows -> open the linked subset -> use the subset as a remote control to enter new values:

If you prefer the other version, you could go to the AddIn folder and change the jsl code.
... or use a code similar to this one:

setting = Is Control Key(); // doesn't work with toolbars :(
dtSubset = Current Data Table() << Subset(
	Linked( Not( setting ) ),
	Suppress formula evaluation( 0 ),
	Selected Rows( 1 ),
	Selected columns only( setting )
);

Is there a trick how to get the CTRL status after triggering the script by a click on a toolbar icon?
Checking-for-modifier-keys-ctrl-alt-shift-in-script

[#9]

Have you thought about row states: Hide & Exclude effectively create a subset for analysis without spawning new (or many) data tables?  This is a no scripting option that will also limit file sprawl.

This would be your alternative method/ except

no need to step 1 "save as", and

instead of step 5/delete  it would be right click hide and exclude (also in columns menu).

Since you mentioned doing this frequently- a few pointers for managing hidden/excluded row states:

• If the selection can be done with column menu/ data filter for "show" and "include" then red triangle "save script to data table" and pick a descriptive name for the subset.  Clear row states when done, and re-run the script when this subset is needed again.

Or- make Row State Columns for each subset...

The Row State type data column can save all row state information including hidden/excluded, but also selected and row marker/color 

Example Below - sample file Big Class with two Row state columns added.

Subset #1: 13 year olds

• Select 13 year olds, right click select matching, right click invert selection
• Right Click Hide/Exclude the non 13's- the Subset of 13 year olds is now ready for analysis.

To preserve and return to this grouping

• Make a new column of Data Type Row State (named RS_13)
• Just for grins set 13 year old row markers to purple squares
• Left Click on the star of RS_13 in columns pane- copy from row states (this sends the current row states to RS_13 column)
• Protip- Column info- engage "Lock" because it is easy to inadvertently over-write

Also did this for a second subset sex=F, Notice how the row 3 selection is part of the saved row state.

•  To deploy, from Row state red star, left click "Copy to Row States". (as illustrated)