turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- Discussions
- :
- Graph builder

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Feb 3, 2009 3:20 AM
(1795 views)

Hi,

I have a couple of questions about the graph builder in JMP 8. I'm a fairly new user of both JMP and this forum, so I apologise if this is the wrong place for it!

I am trying to use the graph builder to do wafer mapping (Fit X by Y) plots. I have two columns with Die X and Die Y positions, a column with a failure bin category (a nominal data set) and a column with a leakage current (a continuous data set). The data is also split into 25 wafers per lot.

I am using different colours to identify individual failure bins, or the leakage current (by grouping the continuous data).

My first question is: when I'm plotting the continuous data (which I put into the overlay area), I can only five different groupings (i.e. colours) in the legend, which are not evenly spaced. Is it possible to control (using JSL ideally) how many levels there are and how the levels are split? This is possible using the contour plot platform and I am currently using that as a stopgap, but the contour plot isn't ideal for this kind of application.

Secondly, when I group by wafer, it defaults to a maximum of four plots horizontally per row (which gives 7 rows with only 1 graph on the last row). Can I tell JMP to plot five graphs per row (which would give me a much more appealing 5 x 5 layout)?

I hope these questions make sense and that someone has some ideas!

Many thanks,

Matthew.

I have a couple of questions about the graph builder in JMP 8. I'm a fairly new user of both JMP and this forum, so I apologise if this is the wrong place for it!

I am trying to use the graph builder to do wafer mapping (Fit X by Y) plots. I have two columns with Die X and Die Y positions, a column with a failure bin category (a nominal data set) and a column with a leakage current (a continuous data set). The data is also split into 25 wafers per lot.

I am using different colours to identify individual failure bins, or the leakage current (by grouping the continuous data).

My first question is: when I'm plotting the continuous data (which I put into the overlay area), I can only five different groupings (i.e. colours) in the legend, which are not evenly spaced. Is it possible to control (using JSL ideally) how many levels there are and how the levels are split? This is possible using the contour plot platform and I am currently using that as a stopgap, but the contour plot isn't ideal for this kind of application.

Secondly, when I group by wafer, it defaults to a maximum of four plots horizontally per row (which gives 7 rows with only 1 graph on the last row). Can I tell JMP to plot five graphs per row (which would give me a much more appealing 5 x 5 layout)?

I hope these questions make sense and that someone has some ideas!

Many thanks,

Matthew.

11 REPLIES

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Feb 4, 2009 6:04 AM
(1489 views)

Hi Matthew,

When you use a continuous variable for grouping, the ranges are determined by quantiles so that about the same number of observations are in each group. When I want more control, I make a new categorical column with just the ranges I want and use that as the grouping/overlay column.

For your other question, we don't have a way to customize the wrapping graphs per row. I will see if 8.0.1 can at least make a better choice for cases like 25.

When you use a continuous variable for grouping, the ranges are determined by quantiles so that about the same number of observations are in each group. When I want more control, I make a new categorical column with just the ranges I want and use that as the grouping/overlay column.

For your other question, we don't have a way to customize the wrapping graphs per row. I will see if 8.0.1 can at least make a better choice for cases like 25.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Feb 4, 2009 9:14 AM
(1489 views)

Hi Xan

thanks for the note. If it could be included in 8.0.1 (and the 64-bit Linux version of JMP 8, which we will be moving to when it's released) that would be great!

Matthew.

thanks for the note. If it could be included in 8.0.1 (and the 64-bit Linux version of JMP 8, which we will be moving to when it's released) that would be great!

Matthew.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 11, 2009 11:22 AM
(1489 views)

Hi --

I am looking at a trial version; and it doesn't look like the ranges are determined by quintiles.

Namely: I have a variable with min=.003; median=.02; 99.5'th quartile=.086; max.=2.4 -- and when I try to use it to wrap the plots (of two other variables) by I am getting three empty plots with ranges [2.4 - 2.4] each.

Is there any way to make it work (other than manually creating rank or quintile variables)?

Thanks!

I am looking at a trial version; and it doesn't look like the ranges are determined by quintiles.

Namely: I have a variable with min=.003; median=.02; 99.5'th quartile=.086; max.=2.4 -- and when I try to use it to wrap the plots (of two other variables) by I am getting three empty plots with ranges [2.4 - 2.4] each.

Is there any way to make it work (other than manually creating rank or quintile variables)?

Thanks!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 23, 2009 10:52 AM
(1489 views)

I created a data table with that shape but am not seeing the problem. I know there was a similar problem which has been fixed for JMP 8.0.2 (coming soon) that was triggered when a large number of rows were excluded (as might happen with the data filter).

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 23, 2009 1:19 PM
(1489 views)

I too have a question about the Graph builder. I used this great feature but then if you try and copy or save the Graph builder as a .doc or .html format it makes it unreadable. Even if I journal it first then save it still is unusable.

I normally do alot of Copying out of JMP directly to an email(Microsoft Outlook). The Graph builder is quick and I prefer using it but without this functionality its unusable to me.

Thanks,

Rod

I normally do alot of Copying out of JMP directly to an email(Microsoft Outlook). The Graph builder is quick and I prefer using it but without this functionality its unusable to me.

Thanks,

Rod

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Sep 24, 2009 5:30 AM
(1489 views)

Do you know about Paste Special? That lets you choose which format of the several that JMP provides. I know of problems with WMF (Windows Metafiles) losing labels, but I thought EMF (Enhanced Metafiles) and bitmaps worked OK.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Oct 8, 2009 2:42 PM
(1489 views)

Can you see the entire script I just posted? I can see it with an edit, but it is truncated on the postings.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Oct 8, 2009 2:49 PM
(1489 views)

the rest of the script..looks like it was clipped...restarted at widx

widx< :WfrIDX[Row() - 1, Empty()], :WfrIDX[Row() - 1, Empty()] + 1));

wrow< wcol<

xwid=90*Col Maximum(wcol)+360;

ywid=90*Col Maximum(wrow)+200;

gb=Graph Builder(

Size( ::xwid, ::ywid ),

Variables(

X( :x ),

Y( :y ),

Group X( :WfrCol ),

Group Y( :WfrRow ),

Overlay( :FMAX_WC )

),

Elements( Points( X, Y, Legend( 8 ) ) ));

//--get listing of wafer numbers by WfrCol and WfrRow and put in teh top left corner

//--first find min x and max y

xp=Col minimum(:x)+1;

yp=Col maximum(:y)-1;

summarize(wtt=by(widx,:wafer));

gb_rpt=report(gb);

gtext= "xx< gtext=munger(gtext,1,"_xp_",char(xp));

gtext=munger(gtext,1,"_yp_",char(yp));

for(i=1,i<=nitems(wtt[1]), i++,

xx=gb_rpt[FrameBox(i)];

cmd=munger(gtext,1,"_wfr_","\!""||wtt[2]*||"\!"");*

eval(parse(cmd))

);

ttl="c:\temp\"||lotid||"_fmax.png";

gb_rpt[PictureBox(1)]< mail("mymail@hotmail.com","lot "||lotid||" Fmax","See attached graph", ttl);

widx<

wrow<

xwid=90*Col Maximum(wcol)+360;

ywid=90*Col Maximum(wrow)+200;

gb=Graph Builder(

Size( ::xwid, ::ywid ),

Variables(

X( :x ),

Y( :y ),

Group X( :WfrCol ),

Group Y( :WfrRow ),

Overlay( :FMAX_WC )

),

Elements( Points( X, Y, Legend( 8 ) ) ));

//--get listing of wafer numbers by WfrCol and WfrRow and put in teh top left corner

//--first find min x and max y

xp=Col minimum(:x)+1;

yp=Col maximum(:y)-1;

summarize(wtt=by(widx,:wafer));

gb_rpt=report(gb);

gtext= "xx<

gtext=munger(gtext,1,"_yp_",char(yp));

for(i=1,i<=nitems(wtt[1]), i++,

xx=gb_rpt[FrameBox(i)];

cmd=munger(gtext,1,"_wfr_","\!""||wtt[2]

eval(parse(cmd))

);

ttl="c:\temp\"||lotid||"_fmax.png";

gb_rpt[PictureBox(1)]<

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Oct 8, 2009 4:39 PM
(1489 views)

gzm, The forum software seems to like to snack on JSL code, but you can circumvent it with around the text. Here's what I see of your code from the email notification:

No way I know of to avoid the loss of indentation.

EDIT: I also added spaces in [ i ] -- otherwise it seemed to still trigger italics.

Message was edited by: xan@jmp

No way I know of to avoid the loss of indentation.

EDIT: I also added spaces in [ i ] -- otherwise it seemed to still trigger italics.

Message was edited by: xan@jmp