topic Re: How do I count defects in a process defined by multiple column means and standard deviations? in Discussions

How do I count defects in a process defined by multiple column means and standard deviations?

Victor60 — Wed, 20 Sep 2023 15:23:22 GMT

Data table is 180K rows. Data is operational recorded data for a steady state process. Four columns measure the pressure in four separate workstreams of the process. Four columns measure the time to create the pressure in the four separate workstreams of the process.

For each workstream, its mean and standard deviation are required, for both pressure and time, to define a threshold.

For each workstream, a count must be registered when pressure is greater than 5X the workstream's standard deviation added to the workstream's mean. Similarly, a count must be registered when time is 5X the workstream's standard deviation added to the workstream's mean.

Finally, if BOTH pressure AND time exceed those workstream limits, a defect must be registered. Pressure AND time must both exceed their individual thresholds to define a defect in this process.

Is there a way to create a script or a process to automate these evaluations, so that for each 180K row data file the final defect count can be made?

Re: How do I count defects in a process defined by multiple column means and standard deviations?

WebDesignesCrow — Mon, 02 Oct 2023 02:11:00 GMT

You have provided very detailed analysis sequence to produce the result that you want.

I think it is possible through scripting.

Maybe you want to test a reduce data set first (say 1000 rows with all those 4 columns) using JMP GUI.

Turn-on the workflow (JMP 17 onwards) while you are doing that.

Tabulate/Summarize (mean / std dev) --> JOIN --> Add count formula column with condition: if time is 5x std + mean --> Add logic formula column (AND)

After you are satisfied with your results, tweak your script for automation.

Regards

Re: How do I count defects in a process defined by multiple column means and standard deviations?

txnelson — Mon, 02 Oct 2023 03:33:34 GMT

Assuming that your first Work Stream column are called "WS1Pressure" and "WS1Time" The following script will create a new column that has the value of 1 when a defect for a given rows is found, and 0 when it is not.

Names Default To Here( 1 );

dt = Current Data Table();
dt << New Column( "WS1 Defect",
	formula(
		As Constant(
			WS1PressureMean = Col Mean( :WS1Pressure );
			WS1PressureSTD = Col Std Dev( :WS1Pressure );
			WS1PressureLimit = WS1PressureMean + 5 * WS1PressureSTD;
			WS1TimeMean = Col Mean( :WS1Time );
			WS1TimeSTD = Col Std Dev( :WS1Time );
			WS1TimeLimit = WS1TimeMean + 5 * WS1TimeSTD;
		);
		If( :WS1Pressure > WS1PressureLimit & :WS1Time > WS1TimeLimit,
			1,
			0
		);
	)
);

If you replicate this code for the other 3 Work Streams, you can then use the Summary or Tabulate platforms to do your summaries of all of the defects found.