XBAR-R Question

[Former Member]

Hi Forum,

I have data summarized and shaped like this (sorry for the lack of formatting, but hopefully you'll get the idea):

StartDateTime EndDateTime Attribute Average Count Max Min Total StdDev

04/02/2009 04:53:33 04/02/2009 05:53:33 Att1 0.77 422.00 1.47 0.66 326.47 0.06

04/02/2009 05:53:33 04/02/2009 06:53:33 Att1 0.76 258.00 1.41 0.63 195.39 0.07

04/02/2009 06:53:33 04/02/2009 07:53:33 Att1 0.78 390.00 1.42 0.64 302.46 0.06

04/02/2009 07:53:33 04/02/2009 08:53:33 Att1 0.77 222.00 1.42 0.61 171.09 0.07

04/02/2009 08:53:33 04/02/2009 09:53:33 Att1 0.77 146.00 1.41 0.65 112.09 0.07

04/02/2009 09:53:33 04/02/2009 10:53:33 Att1 0.78 180.00 1.01 0.61 139.61 0.06

The sample sizes are evidently not consistent, it is in-line process monitoring, grouped by 1 hour time interval.

I am trying to plot this data on an XBAR-RANGE chart. I can plot the XBAR part OK by simply setting the Value column = "Average" and SampleSize = "Count". This accurately generates the control limits as well.

The problem is that, as is, the RANGE is always 0 (because the subgroup size is 1). I however KNOW the range ("Max" - "Min") but there isn't anywhere to enter it on the display template. Am I missing something in the template or can anyone offer any advice?

Thanks.

[Former Member]

In the current SPC implementation, Range and other statistics are calculated. Also, the control limits are probably not being calculated correctly if the "true" subgroup size was not actually 1.

[agentry_src]

Hi David,

Don't aggregate your data points before plotting them. But preprocess your data and add a column to the result set for each increment of time that you are interested in. Then use the new column for your subgrouping. That should get you the range (since range does require multiple points) for your chart.

If anyone else wants to chime in, please do so. This is not my best area of expertise.

Good luck,

Mike

[Former Member]

Hi Mike,

Conceptually, that makes sense, but there isn't a field to enter a subgroup size column (like there is for sample size). Maybe I'm missing it or it is labeled something different?

Thanks again,

Dave

[Former Member]

Also, in general, control limits are NEVER calculated in a real production environment when you're trying to "manage" variability. They are generally only calculated when you're trying to "learn" the process variability. Once the process variability is understood, you generally fix your control limits at certain values and manage the process to those limits. Only when there is a fundamental change to the process (new equipment, procedures, materials) would you normally revisit the control limits.