210505_en-US

liliconcom

process screening for discrete data

My data consists of the number of defects per quarter for different processes. I have included a sample of my data below. I have about 800 different processes. I want to be able to screen all of them, and flag the ones that had the most variablity over the past 12 quarters.

 

- I have used the U control chart builder but that becomes a more manuel process checking through 800 processes  

 

process  quarter-year Number of defects

process #1 1-2015     0

process #1 2-2015     0

process #1 3-2015     1

process #1 4-2015     0

process #1 1-2016     1

process #1 2-2016     1

process #1 3-2016     0

process #1 4-2016     3

process #1 1-2017     3

process #1 2-2017    11

process #1 3-2017     6

process #1 4-2017     2

process #2 1-2015     0

process #2 2-2015     0

process #2 3-2015     0

process #2 4-2015     0

process #2 1-2016     1

process #2 2-2016     1

process #2 3-2016     1

process #2 4-2016     2

process #2 1-2017     2

process #2 2-2017    1

process #2 3-2017     2

process #2 4-2017     2

Re: process screening for discrete data

Have you tried using Partitioning platform, that will give you an easy way of determining which processes are associated with higher defects https://www.jmp.com/support/help/14-2/partition-models.shtml

Re: process screening for discrete dataI want to be able to monitor the processes over the quarters. For example, if a process has 0 defects for the first 11 quarters, and has 12 defects on the 12th quarter- I want to be able to see that. Not necessarily which ones are associated with higher defects.Re: process screening for discrete data

Assuming you have a column of data containg the date of each batch you might try using the quarter and year functions to derive new variables and then summarise total number of defects by year and quarter using tabulate or graph builder