search

luca-scr / qcc

An R package for Quality Control Charts
https://luca-scr.github.io/qcc/
43 stars 17 forks source link

Standard deviation value #41

Closed mantikera closed 1 year ago

mantikera commented 1 year ago

Using the qcc package I've noticed a difference between the value of standard deviation value that is showed in your plots and the value that I calculate in other softwares even in Excel. Do you know why there is this difference? Thank you. sd_difference

data peso456.csv

birdybro commented 1 year ago

I wonder if the behavior is the same prior to this commit --> https://github.com/luca-scr/qcc/commit/abc75f55e9fd0efee40ff966c13b22d8631af6ed or not?

mantikera commented 1 year ago

Hi @birdybro sincerely I don't know. I'm using the latest qcc's package version but the difference between sd values persists. I will check it out. Thanks.

birdybro commented 1 year ago

I was curious, so I talked to someone who knows a lot more about statistics and quality engineering than I do (we are also considering using this library along with R to explore other options). He uses Excel, Minitab, and JMP (another SAS similar to Minitab) for this kind of stuff, and he said that JMP also varies from Excel's std dev, so I would assume that as long as your end result (the plot shape) is the same it shouldn't be an issue.

luca-scr commented 1 year ago

I have explained several times that std deviation is computed using functions sd.xbar(), sd.xbar.one(), sd.p(), etc. depending on the chart type. Please see the help pages that explain the method used, or look at the source code. Please note that all these compute the process standard deviation which takes into account the within-rationale-group std deviation. And this is not the same as the whole sample std deviation. If you are not happy with the default value you may provide another value using the optional argument std.dev available in function processCapability().

mantikera commented 1 year ago

Thank you @luca-scr for your reply. I'm very happy with the results and as I said your package is awesome. I'm just curious and I noticed the difference because I'm Minitab user and SAS user as well but I use RStudio and your package as an alternative for customers that doesn't want to buy a software or can't afford a software for statistical analysis. So the customer uses a free one. And your package apllication is terrific for my purposes and the customer's purposes. I just want to know the difference to explain to them in case someone ask for. Thank you very much.

mantikera commented 1 year ago

Thank you @birdybro. I'm a Minitab user too and I use SAS sometimes. I understant the difference of values from Excel (that I don't use for this purpose) but as I said I noticed the variation in results and want to know why this happens. Usually I do analysis in RStudio/R's package and confirm using Minitab to check the results. If are equal it's Ok. I can proceed with my work. If not I try to discover why the values don't match. That's it.