P Control Chart ( wrong value plotted with Local Data Filter on when NG counts is zero)

[#1]

Dear JMP communities, 

I am having a very strange problem when trying to use P Control Chart as illustrated below:

On 02/24 02/25, 02/26; at Station 2, NG counts are Zero, but the P Chart gave a non-zero value; it turned out the local Data Filter didn't work in this case, it included the Station 1 NG counts. For example on 02/25, (13+0)/(258+13)=0.048.

But when NG counts is not zero (02/27), the local data filter won't include the Station 1 results. Or when there is no Station 2 results  (03/10), it also won't plot station 1 results. 

Really appreciate if anyone provide some insights on how to solve this strange problem.  Raw JMP file also provided. 

[#2]

Hi Wenye,

I tried to plot your data and had the same feeling that the local data filter is not work properly under the control chart builder platform (it seems always grouping the count by date regardless of the local data filter - only works when there is no overlapping on date and station).

Alternative way is to build your control chart by station (maybe this is JMP assumption: control chart should be built per station), it seems working that way:

Hope this helps.

Best,

Harry

[#3 · Accepted solution]

You need to turn off Show Excluded Region.  Try the following code:

Control Chart Builder(
	Show Excluded Region( 0 ),
	Class( Shewhart Attribute ),
	Variables(
		Subgroup( :DATE ),
		Y( :NG Counts ),
		n Trials( :Total Inspected Counts )
	),
	Chart( Points( Statistic( "Proportion" ) ), Limits( Sigma( "Binomial" ) ) ),
	Local Data Filter(
		Width( 172 ),
		Add Filter(
			columns( :INSPECTION_STATION ),
			Where( :INSPECTION_STATION == "Station 2" ),
			Display( :INSPECTION_STATION, N Items( 2 ), "List Display" )
		)
	)
)

This will give you

[#4]

Thank you very much Tonya, I spent many hours on it, couldn't find this "magic" option.