topic Re: carte de contrôle avec des données qui sont déjà des moyennes in Discussions

carte de contrôle avec des données qui sont déjà des moyennes

Francis38 — Wed, 15 Nov 2023 15:50:49 GMT

Bonjour

je cherche à indiquer à la plate forme qualité que les données que je récolte sont déjà des moyennes (de 9-21-49 valeurs selon les cas).

je n'ai pas trouvé où indiquer à JMP que les valeurs de la table sont déjà des moyennes. Du coup, les limites calculées sont fausses par rapport à ce que je calcule pour mes données sous Matlab.

Un indice de la bonne case à cocher ?

merci d'avance.

Re: carte de contrôle avec des données qui sont déjà des moyennes

statman — Wed, 15 Nov 2023 16:08:25 GMT

Welcome to the community. I'm a little confused by the question. Do you have all of the data or only the averages? If you do not have the complete data set, it is impossible to calculate the range of the within subgroup variation. This is required as the average range is used to calculate control limits for both the range chart and the average chart. Another option is to do an IR chart (Individual moving range). The IR chart is not answering the same question as the X-bar chart.

Re: carte de contrôle avec des données qui sont déjà des moyennes

Francis38 — Thu, 16 Nov 2023 07:06:38 GMT

Comme indiqué, je n'ai que des moyennes.

j'ai bien vu les possibilité offertes si on dispose de l'ensemble des données, mais ce n'est pas mon cas.

cordialement

Re: carte de contrôle avec des données qui sont déjà des moyennes

WebDesignesCrow — Tue, 21 Nov 2023 01:51:29 GMT

When you mentioned the calculated limit not same like in Matlab, what type of method that you used in Matlab to calculate control limit?
If you want a flexible calculation of control limit (depending on your scenario), maybe you can use Control Chart Builder.
If you just want to add a note to state that the column is average, you can define the column properties & add notes that it is an average ( but i doubt it will change the calculation ).
For single data point (in this case “average value”), you can either use IR method or Levey-Jennings in JMP
IR in this case will treat your average as single point & sigma is moving range between the 2 averages.
Levey-Jennings will treat you average as single point as well & sigma is Levey-Jennings (overall standard deviation of all averages).

p/s: Earlier post mentioned Minitab instead of Matlab. Thanks @ih for the correction.

Re: carte de contrôle avec des données qui sont déjà des moyennes

ih — Tue, 21 Nov 2023 23:08:49 GMT

I am no expert here but I second the question from @WebDesignesCrow about how this is done in MATLAB. The variation of individual samples seems like it would be central to subsequent analyses so I suspect there is some assumption made in MATLAB. Consider the samples data in the example script below. Using the same averages for each group but different variations within the group, the control limits change significantly.

Run this script to recreate the sample data and charts above:

Names default to here(1);

Random Reset(84);

dtRaw = New Table( "Raw Data",
	Add Rows( 20 ),
	New Column( "Raw Data",
		Numeric,
		"Continuous",
		Format( "Best", 12 ),
		Formula( Random Normal() )
	),
	New Column( "Group",
		Numeric,
		"Continuous",
		Format( "Best", 12 ),
		Formula( Floor( (Row() - 1) / 4 ) ),
		Set Selected
	)
);

dtRaw << Run Formulas;

dtRaw:Raw Data << Delete Formula;
dtRaw << New Column("0 Variation", Numeric, "Continuous", Format("Best", 12), Formula(Col Mean(:Raw Data, :Group)));
dtRaw << New Column("Mean", Numeric, "Continuous", Format("Best", 12), Formula(Col Mean(:Raw Data, :Group)));
dtRaw << New Column("StDev", Numeric, "Continuous", Format("Best", 12), Formula(Col Std Dev(:Raw Data, :Group)));
dtRaw << New Column("Half Variation", Numeric, "Continuous", Format("Best", 12), Formula((:Raw Data + :Mean) / 2));

dtSum = (tab = dtRaw  << Tabulate(
	Show Control Panel( 0 ),
	Add Table(
		Column Table( Analysis Columns( :Raw Data ), Statistics( Mean ) ),
		Row Table( Grouping Columns( :Group ) )
	)
)) << Make Into Data Table;
tab << Close Window;



New Window("Control Charts",
	vlb = V List Box(
		dtRaw << Control Chart Builder(
			Size( 570, 150 ),
			Variables( Subgroup( :Group ), Y( :Raw Data ) ),
			Show Control Panel( 0 )
		),
		dtRaw << Control Chart Builder(
			Size( 570, 150 ),
			Variables( Subgroup( :Group ), Y( :Half Variation ) ),
			Show Control Panel( 0 )
		),
		dtRaw << Control Chart Builder(
			Size( 570, 150 ),
			Variables( Subgroup( :Group ), Y( :Mean ) ),
			Show Control Panel( 0 )
		),
		dtSum << Control Chart Builder(
			Size( 570, 200 ),
			Variables( Y( :"Mean(Raw Data)"n ) ),
			Show Control Panel( 0 )
		)
	)
);

ob = (vlb << XPath("//OutlineBox[text()='Control Chart Builder']"));
ob[1] << Set Title("Raw (Random) data");
ob[2] << Set Title("Variation cut in half");
ob[3] << Set Title("No variation - average repeated");
ob[4] << Set Title("Individual values as averages");