Transform variable data into attribute data

Kid006

Starting to get Involved
#1
Greetings
In a recent audit we had a finding regarding a validation in a die cutting press machine, the person incharge of this validation treated the data as variables since we have tolerances in drawings however the output of the press was always the same since we are using a die cut.
The cpk was calculated using the minimum required sample size of 32, for obvious reasons this data throws a non normal distribution and high cpk (this was the finding saying that its not calculated correctly)
Am i able to change the method into a go/no-go (attribute data) increasing the sample size? Our product doest not require a high precision measurement instrument and it is measure with tape.
All die cuts are being verified yearly and the go-no go will be the product inside the spec.
Also i would to include in the validation that since we cant certantly know if the die cut throws variation in the product because we dont have an optical comparator or such kind of instrument it will be treated as attribute?

Any comment will be appreciate.
Thanks
 
Elsmar Forum Sponsor

Bev D

Heretical Statistician
Staff member
Super Moderator
#2
Who said it wasn't calculated correctly? what qualifications do they have to make that assessment? I see nothing wrong with that calculation, in fact moving to an attribute calculation is actually contrary to the purpose of the Cpk calculation.
 

Miner

Forum Moderator
Staff member
Admin
#3
I do not recommend treating variable data as if it were attribute. You would be throwing away a lot of information and forcing yourself to use very large sample sizes. I assume that the auditor had an issue with you using nonnormal data to calculate capability without taking its shape into consideration. Given that, there are several ways to handle non-normal data.
  • Perform a non-normal capability study
  • Transform the data before performing a capability study
  • Use the nonparametric capability approach 1590692296587.png
However, make sure you have a stable process without mixed process streams before applying any of these.
 
Last edited:

Kid006

Starting to get Involved
#4
Who said it wasn't calculated correctly? what qualifications do they have to make that assessment? I see nothing wrong with that calculation, in fact moving to an attribute calculation is actually contrary to the purpose of the Cpk calculation.
the auditor said; cpk
talking with the auditor she said; it was not calculated correctly since CPK>11.0 at low,nominal and high parameters, the normal prob plot were vertical straight lines in all cases meaning its not normal and normal process capability report was used for all cases.
 

Miner

Forum Moderator
Staff member
Admin
#5
The only way to get a vertical straight probability plot is if all values were identical. If that is the case, how good is your measurement system?
 

Kid006

Starting to get Involved
#6
I do not recommend treating variable data as if it were attribute. You would be throwing away a lot of information and forcing yourself to use very large sample sizes. I assume that the auditor had an issue with you using nonnormal data to calculate capability without taking its shape into consideration. Given that, there are several ways to handle non-normal data.
  • Perform a non-normal capability study
  • Transform the data before performing a capability study
  • Use the nonparametric capability approach
However, make sure you have a stable process without mixed process streams before applying any of these.
I didnt carry out the validation however i have to followup with the finding
Big samples size are not a problem actually thats why i think into transform the variable data into attribute,
my other idea is tranforming the data with a boxcon transform to have a more realistic cpk
 

Kid006

Starting to get Involved
#7
The only way to get a vertical straight probability plot is if all values were identical. If that is the case, how good is your measurement system?
Our instrument system is not capable to find the variation after the cutting press process, the product is measured with tape and we dont have an optical comparator or any other instrument.
 

Bev D

Heretical Statistician
Staff member
Super Moderator
#8
The only way to get a vertical straight probability plot is if all values were identical. If that is the case, how good is your measurement system?
Given a tape measure and a die cut a vertically straight line is to be expected. The fact that the distribution is not normal is to be expected. The fact that - right now given the proscribed time frame and sample size - the Cpk is large and WILL NOT SUBSTANTIALLY CHANGE no matter what approach the OP takes is enough to say: move on. The process is capable; torturing the data and the organization any more is waste. The auditor is clearly not knowlegable in statistics - they are reacting by rote. At some point common sense and knowledge of physics must be heard.
 

Kid006

Starting to get Involved
#9
Given a tape measure and a die cut a vertically straight line is to be expected. The fact that the distribution is not normal is to be expected. The fact that - right now given the proscribed time frame and sample size - the Cpk is large and WILL NOT SUBSTANTIALLY CHANGE no matter what approach the OP takes is enough to say: move on. The process is capable; torturing the data and the organization any more is waste. The auditor is clearly not knowlegable in statistics - they are reacting by rote. At some point common sense and knowledge of physics must be heard.
These are facts but i cant tell the auditor these points.

Also looking at the calculated cpk in minitab it is clear that the results were manipulated since minitab can't give a solution when there is not enough variation among the sample size.
thats why im trying to change the data to attribute
 

toniriazor

Involved In Discussions
#10
I prefer to rely always on variable, because it gives you a lot more information about your process than go no go gauge. The fact that an auditor has said something does not necessarily means he/she is right. Check how the data was collected and measured. Often there is a lot of copy-paste due to time restraints, lack of resources and etc.
 
Thread starter Similar threads Forum Replies Date
L Cpk after Data Transformation - How to Transform LCL /UCL to get Capability Report Capability, Accuracy and Stability - Processes, Machines, etc. 8
optomist1 Data Analysis - Transform Data or Not Capability, Accuracy and Stability - Processes, Machines, etc. 9
S Surface Finish - Conversion factor to transform Rz Jis in Rz Inspection, Prints (Drawings), Testing, Sampling and Related Topics 6
B To detect the non standard variable IATF 16949 - Automotive Quality Systems Standard 7
J Sample size for creating a data base as a reference to a tested variable Other Medical Device and Orthopedic Related Topics 6
D Do I need part variation while doing Destructive Variable Gage R&R MSA study Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 19
E Sample size for design verification of variable in single use device Design and Development of Products and Processes 20
N Variable Data Test Never Fails Capability, Accuracy and Stability - Processes, Machines, etc. 9
B Variable Gage R&R - Two operators instead of three Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 4
Q Can there be a mix of good and bad parts for a variable GR&R? Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 7
E GR&R for a test fixture that outputs variable data per measurement Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 1
L Gage Blocks: Attribute or Variable Gage? General Measurement Device and Calibration Topics 3
T Attribute vs Variable Inspection Data ISO 13485 Requirement ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 2
A Taguchi Minitab - Continuous Data - What should I choose as response variable? Using Minitab Software 3
BHobbs_Busche All Measurements taken with a Variable Gage have to be Recorded 100% IATF 16949 - Automotive Quality Systems Standard 14
V Why Gauge R&R 10% is acceptable for variable instruments ? Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 5
Douglas E. Purdy C=0 Sampling for Variable Characteristics Quality Manager and Management Related Issues 14
J Passing a variable to the ODBC call in Minitab Using Minitab Software 8
S How to use Dummy Variable in Regression Model in Minitab Six Sigma 2
T Variable Dimensions / Part of Production / PPAP Dimensional Report Inspection, Prints (Drawings), Testing, Sampling and Related Topics 7
L Do all non-random patterns apply to attributes and variable charts including R chart? Statistical Analysis Tools, Techniques and SPC 5
M Sampling for Variable Data to determine the size of a Mould Inspection, Prints (Drawings), Testing, Sampling and Related Topics 1
B Regression analysis with discrete dependent variable and continuous independent var. Statistical Analysis Tools, Techniques and SPC 13
K How to group 5 questions to be one variable for Likert Scale question Using Minitab Software 11
M Destructive VTMV Subgroups - A Destructive Variable Test Method to Validate Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 11
B Using a File Name as a Variable in an Excel Macro Excel .xls Spreadsheet Templates and Tools 11
S Strange PT Ratio on an Excel spreadsheet for variable GRR ANOVA Method Excel .xls Spreadsheet Templates and Tools 2
A Sample Size Selection for Peel Strength for Variable Data using a 95%/99% Confidence Inspection, Prints (Drawings), Testing, Sampling and Related Topics 8
S Probability Density Function of a Function of a Random Variable Statistical Analysis Tools, Techniques and SPC 13
Wayne User's Manuals for Variable Screw Thread Measurement Systems General Measurement Device and Calibration Topics 1
E Gage Repair Shops near Pheonix, AZ - Variable Gages - Micrometers, Calipers, etc. Calibration and Metrology Software and Hardware 3
S What should I use for Variable Sample with Size 3, 4, and 5 - xbar-R or xbar-S? Statistical Analysis Tools, Techniques and SPC 4
X Fanuc Injection Moulding Machine - Variable stage speed/pressure injection profiles Manufacturing and Related Processes 1
M Gage R&R for Big Parts - Approving a Variable Gage using only 1 Part Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 7
K DOE - Variable and Discrete Data - Minitab Using Minitab Software 11
M Regression Analysis Help - Factors influencing a Dependent Variable Using Minitab Software 7
Marc Steve Durnin's Infinitely Variable Transmission World News 0
N Variable Control Limits Calculation - P Chart Statistical Analysis Tools, Techniques and SPC 8
V Questions on Variable and Attribute MSA Interpretation Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 1
N AQL - Sampling plan (variable, inspection level ii, normal inspection S-method) AQL - Acceptable Quality Level 2
A Acceptable Criteria for Variable Gage R&R according to the MSA Manual 3rd Edition Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 3
E Variable Data - Total Percent Nonconforming from Different Nonconformances Quality Tools, Improvement and Analysis 4
E Barcode Format Question - Variable vs. Fixed Segment Length Pharmaceuticals (21 CFR Part 210, 21 CFR Part 211 and related Regulations) 3
K Complying with Ford customer requirement 4.35 Variable Gage Studies General Measurement Device and Calibration Topics 3
J Critical Characteristics need to be measured 100% if they are variable in nature Inspection, Prints (Drawings), Testing, Sampling and Related Topics 5
bobdoering Determining a Good Variable for Small Scale Weld Integrity Capability Evaluation Capability, Accuracy and Stability - Processes, Machines, etc. 7
P P-Chart with variable sample size - How to update the chart? Statistical Analysis Tools, Techniques and SPC 6
A Sampling - Measuring and Inspecting a part with both Variable & Attribute Data Inspection, Prints (Drawings), Testing, Sampling and Related Topics 12
C Minitab Column Sorting Issue - Invalid variable(s) to sort Using Minitab Software 9
R Sample determination using minitab if everything elses is a variable Inspection, Prints (Drawings), Testing, Sampling and Related Topics 3

Similar threads

Top Bottom