D21977: One Way Anova Test

Devanshu Agarwal noreply at phabricator.kde.org
Sat Jun 22 20:19:26 BST 2019

devanshuagarwal marked 5 inline comments as done.
devanshuagarwal added inline comments.


> sgerlach wrote in HypothesisTest.cpp:476
> do we have global variables? If yes, why do we need them?
> Avoiding global vars is a good programming practise

This function clears variables in HypothesisTestView, so that new results can be added. We dont actually have that many global variables.

> sgerlach wrote in HypothesisTest.cpp:506
> how efficient is qPow()? Does it make sense to use a gsl function for small inter powers to get better performance? See https://www.gnu.org/software/gsl/doc/html/math.html#small-integer-powers

Thanks :)

> sgerlach wrote in HypothesisTest.cpp:978
> this is only correct, when the distribution is symmetric? Is this true for the t-distribution?

Yes, firstly I used gsl_cdf_tdis_P(value, df) + gsl_cdf_tdis_P(-value, df). but the results from JASP and online calculator are not matching with this. The results are matching with 2*gsl_cdf_tdis_P(value, df)

> sgerlach wrote in HypothesisTest.cpp:994
> not gsl_cdf_gaussian_P?

yes it should be gsl_cdf_gaussian_P. Actually, I have left ztest backend for now, I am not able to find online calculator and neither ztest is included in JASP. So for time being, I am not concentrating on it.

> sgerlach wrote in HypothesisTestDock.cpp:253
> please use cbTest->currentItem() and a matching enum (see above)

Is this not correct or less efficient?

> sgerlach wrote in HypothesisTestDock.h:77
> can we collect all the bool vars in an enum or a bit-field. Do we even need all of them (one_way=false is equal to two_way=true and vice-versa)?

No, technically we dont need separate variables, but it avoids confusion and also gives us the clear overview.

I will think about using bit-field, but again, it will make things complex and will increase bugs.

  R262 LabPlot


To: devanshuagarwal, sgerlach, asemke
Cc: kde-edu, #labplot, narvaez, apol
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-edu/attachments/20190622/6b757fe4/attachment-0001.html>

More information about the kde-edu mailing list