Apr 26, 2012
Report by Exception using IR Control Charts
Reporting by exception may be just a little counterintuitive for checking control charts?
A key concept of control chart review is that a person actually looks at each chart. This however
gets a little tedious with the daunting task of frequently reviewing hundreds of process variables
that typically contain few special cause events. For reviewing many control charts at once it may
be beneficial to only present charts with control chart test alarms. In addition we typically
review ranges of 25 to 250 data points where only the most recently added data is interesting.
Test alarms early in the reporting period are just noise have typically been addressed in earlier
reports and no longer need attention.
This script is design to simplify this reporting task. Two sets of control charts are generated.
The first contains control charts for all the variables. The second only contains reports
with the specified test alarms for the specified reporting period. The reporting period can be
either a range of rows, or it can be for a certain number of measurements from the end of the table.
*The data for this script must be sorted sequentially by measurement order.
*This script is designed for IR control charts only. If you give it labels that are not unique,
an Xbar and R chart dialog will launch. (no error control for this)
*The Ys (process variables) must be numeric continuous data.
*The Label must be unique for each row.
*Test 1 to 4 are implemented as options (feel free to add more)
In the column dialog, specify continuous numeric columns for control charts.
Select the control chart alarm 1 to 4 (test 1 is on by default).
Enter either the beginning and ending row of the range you want alarms from or
the number of days from the end of the last row. Click OK or Cancel.
This script assembles control chart functions with the extensive use of "Insert Into" and "Substitute Into"
This script was a lot of fun to put together but now JMP has the Process Screening platform which does all this and more, plus it doesn't require scripting. Before trying to make this script work for you, please take a look at the new platform. Also, the new platform is scriptable which essentially gives you a whole new set of complex functions without all the trouble. For example, adding control limits to your process data table from a control chart limits table. This is so much easier to do by scripting the platform rather than DIY. (check out the scripting index for examples)