Subscribe Bookmark RSS Feed

Bivariate Chart changes even if Automatic Recalc is not selected, how to stop this please?

mattpace0

Occasional Contributor

Joined:

Aug 23, 2016

The Automatic Recalc is not selected, but the chart changes with the row filter.

Is this deliberate, if so, how may it be disabled please?

 

Capture1.PNGCapture2.PNG

18 REPLIES
txnelson

Super User

Joined:

Jun 22, 2012

Autorecalc controls the recalculation of statistics and calculated elements displayed on a graph, but display single value elements(row values) will appear and go away when the "Show" value is selected in the data filter.  Notice, in your screen display, that the statistical values in the Summary of Fit display table do not change, just the displayed single data points.

Jim
mattpace0

Occasional Contributor

Joined:

Aug 23, 2016

Why does only the Bivariate chart change and not others?

Using automation to generate many charts, each based on a filter, I'd like the charts to stay as they were when generated, as the variability chart does

 

Capture1.PNGCapture2.PNG

txnelson

Super User

Joined:

Jun 22, 2012

The key to this is the statement from my original response.....and it is a fine point, so I can understand why you didn't pick up on it
"Autorecalc controls the recalculation of statistics and calculated elements displayed on a graph"
What is displayed in the Variability chart is a "Mean". A "Mean" has to be calculated. So if Autorecalc isn't turned on, the displayed "Mean" will not be recalculated.
Jim
mattpace0

Occasional Contributor

Joined:

Aug 23, 2016

I am sorry, I am not communicating.

 

Both charts are created from the same data set.  The Mean is a column of numbers, nothing is calculated

The first image one item from each filter is selected and the charts generated.

Capture1.PNG

 

 

In the second image, an addition item was selected from the filter.  The additional data is added to the Bivariate chart, but not the Variability chart.

Capture2.PNG

 

 

In the third image, a different item is added to the filter and the added data points do not appear on either chart.

Capture3.PNG

 

In the forth imahe, Autorecalc was selected on the Variability chart and the additional data appears on it.

Capture4.PNG

 

 

 

The Bivariate chart appears to act differently than the Variability chart when the filter is changed.  

 

I am running an automation script I created that generates many charts based on selected filter items.

The fifth image is the result of running the script.  It generated one Variability chart for each filter item "ROI"  The charts do not change after the filter is clearedd from the data table.  This is the behavior I was expecting from the Bivariate chart, but when the filter is cleared the Bivariate chart plots all data.

Capture5.PNG

 

The sixth image is a Bivariate chart after the filter is cleared.

Capture6.PNG

 

 

 

 

ryan_dewitt

Community Manager

Joined:

Jun 26, 2014

Data Filters can be added either globally or locally to each platform and associated reports. However, sometimes we may not know where and when to filter.  Here's a video that explains the difference and a variety of applications in JMP.  Hope this helps. 

  • Data Filter: Every JMP report using that Data Table is filtered.
  • Local Data Filter: Exclusive filtering in that platform or report. 

 

 

Ryan DeWitt
Community Manager | Learning Content Specialist | SAS JMP Customer Care
billw_jmp

Staff

Joined:

Jul 2, 2014

In your data filter you have Show and Include checked.  That causes all the points that are not part of your selection to excluded and hidden.  If you want those points to still be shown no matter your choice(s) you will need to uncheck those boxes.  The points that are not part of the selection will be grayed and points of choice will be darker(color dependent)

mattpace0

Occasional Contributor

Joined:

Aug 23, 2016

The real question is why does the Bivariate chart act differently from the variability chart

txnelson

Super User

Joined:

Jun 22, 2012

Here is my guess as to what you are seeing.  You are running the Bivariate.  You then open a data filter and make some selections.  You then run the Variability Chart.  You then go back to the filter and make changes.  You see the changes in the Bivariate display, but not in the Variability display.

 

The explanation:

The Bivariate chart is not calculating anything, and therefore it's display is changing as you change the filter.

The Variability chart was generated with a filter turned on.  The Variability chart is a calculated environment, not just a display environment.  So the categories on the X axis were determined during a filtered situation.  When you change the filter, if new raw data are now available to the Variability chart, the chart does not know what to do with the data, since the parameters of the chart were determined without those data points.

 

Below is a simple script that illustrates the situation.  Once the bivariate runs and the variability chart comes up, go to the data filter and select "clear"  what you will see is that even though more data values appear on the bivariate chart, the variability chart does not know how to handle the new points.  Now go to the data filter and select the Males.  All data points on the variability chart go away, but only half of the values on the bivariate chart. 

dt=open("$SAMPLE_DATA/big class.jmp");
Oneway( Y( :height ), X( :age ) );
wait(0);

dt << Data Filter(
	Location( {635, 44} ),
	Mode( Show( 1 ), Include( 1 ) ),
	Add Filter( columns( :sex ), Where( :sex == "F" ) )
);
wait(5);

dt <<Variability Chart(
	Y( :height ),
	X( :age ),
	Max Iter( 100 ),
	Conv Limit( 0.00000001 ),
	Number Integration Abscissas( 128 ),
	Number Function Evals( 65536 ),
	Analysis Type( "Choose best analysis (EMS REML Bayesian)" ),
	Show Range Bars( 0 ),
	Show Separators( 0 ),
	Std Dev Chart( 1 ),
	SendToReport(
		Dispatch(
			{"Variability Chart for height"},
			"Variability Chart",
			FrameBox,
			{Grid Line Order( 2 ), Reference Line Order( 3 )}
		)
	)
);
Jim
mattpace0

Occasional Contributor

Joined:

Aug 23, 2016

Thank you.  I believe the key is as you stated, is the difference between a calculating and non-calculating chart or environment.  I have automated distribution generation also and it works like variability, which would seem consistant.

Is there a way to stop this behavior?  The automation process generates many charts, each based on a unique filter, but this is defeated if they all look the same in the end.