QI Macro Histogram - Can someone define *sorted data*?

[qcman]

Can someone define *sorted data* as it applies to the screen shot of a histogram I ran. This occurred on one of the 15 different dimensions consisting of 30 readings each on an extrusion profile.

 

[reynald]

The conclusion given by the Excel macro comment is wrong. Cpk>Ppk does not mean your process is unstable.
So I guess you don't have to trust that "sorted" data comment as well. I personally never heard of that term in a capability analysis.

 

[qcman]

Thank you for confirming my suspicions. Still kind of wondering what sorted data means. Sounds like I chose certain data as in hand picked (?)

 

[Miner]

The comment regarding sorted data actually does make sense. If the data were sorted low to high (or high to low) the moving or subgroup ranges would be different than they would have been had they been in time series order. This could cause some unusual results when calculating capability indices.

Can you attach your data? Or at least the time series plot of the data?

Also, I noticed this is extrusion data. What type of extrusion product? There is a good chance that your data is also auto correlated, which can also influence control limits and capability indices.

 

[qcman]

Attached are the 15 readings taken for dim #2 on a 100pc aluminum extrusion run. Data collector is not here this week so I can not confirm if the readings are random. Based on the numbers I received and knowing the process I would say they are sequential parts.

 

[Miner]

The data appear to be in time series order, and there is some instability in the process as can be seen from the attachment. There was some instability in subgroups 4 - 12, then the process appears to have stabilized. After removing subgroups 4 - 12, the data were normal.

 

[qcman]

Miner you are correct That dip did not register as the cause with me.

 

[reynald]

The comment regarding sorted data actually does make sense. If the data were sorted low to high (or high to low) the moving or subgroup ranges would be different than they would have been had they been in time series order. This could cause some unusual results when calculating capability indices.
.

Impressive! I have not imagined that possibility. Thanks a lot. Always a great feel to learn something new.