Non-normal Capability Analysis for Multiple Variables: Illustration with Practical Example

[@generalmanageroperations6387]

well explained

[@learnandapply]

Thank you for your valuable comments.

[@tamilsiva1947]

Thank you sir.

[@learnandapply]

Thank you for your valuable comments... Please share it with your friends and colleagues..

[@mahmoudelemam5341]

I f i have to check tube in many points in it is out side diameter say 7 points (7 out side diameters measurements).how i can calculate process capability and i have to check other output like internal diameter in same manner in 7 points i have to consider the internal and external each one is out put separately.and i have to check also thickness in 7 points and runt out in same points .all this dimension are in one part. I have 20 pieces to produce in serial production and after that continue to produce full serial production . I can this as sample to do analysis.

[@tamilsiva1947]

3 parameter weibull has 0.5 p value and box cox transformation has 0.351 p value. Then why we choose largest extreme p value. Its only 0.25 ! Please explain sir

[@learnandapply]

If you have no problem in changing data values by transforming it, then you can use transformation option. In case of weibull it is with LRT p value. Values with LRT P need to considered at the last i.e. if no p-value is more than 0.05. This is calculated by adding additional parameter to fit the distribution.

[@tamilsiva1947]

@@learnandapply thank you for your explanation sir.

[@learnandapply]

Your welcome 😊

[@tamilsiva1947]

Sir, how to choose goodness of fit values? Which value to be cosider? I am confused in that!

[@learnandapply]

Thank you for your valuable comments.. In selection of best fit, null hypothesis is opposite. I don't want to make it complicated. Simply, the distribution having highest p-value will be the best fit distribution. I had mentioned it in each video wherever required.

[@niakusnaini1543]

Sir my data was distribution non normal and then i try to find out what distribution that fit with my data. But when i use distribution identification there was not distribution that show p value > 0.05 . What i have to do with my data sir?

[@learnandapply]

Please go to LRT p values (Likelihood Ratio Test) in the last column and select a distribution with a p value less than 0.05. Here is a reverse thing, we need to look for a p-value less than 0.05. Please let me know if it helps you 😊.

[@niakusnaini1543]

@@learnandapply very very gratefull for your respons hankyou sir, i will try it🙏🏻 . an then there is something else that i want to make sure, I want to describe the condition of my data, my data is the weight data of a bottle taken from the supplier's production. where a machine has several prints and the data is taken from several different manufacturing days, sampling with composite methode. With data like that, can I still use tools quality analysis or should i use tools quality for multiple variables?

[@learnandapply]

The best option will be one-factor experiments like the Latin square method. You can also use quality tools with stratification.