commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Phil Steitz" <p...@steitz.com>
Subject RE: [math] TTestImpl
Date Mon, 17 May 2004 20:07:51 GMT
Joel,
 
Thanks for the feedback.
>
>1) Is there a specific reason why ttest would throw an exception if there
>is not sufficient data. In my case I would like to execute it with only 2
>data (even if it does not mean anything). The developer should be smart
>enough to make this test BEFORE calling the function if it really needed

We have debated this kind of issue quite a bit.  When a mathematical computation is meaningless
(bad or insufficient data to compute a meaningful result), we have essentially three choices:
1) throw an exception (what TTestImpl does, used when computation is meaningless) 2) return
NaN (when led there by IEE754) or 3) return nonsense.   We have not been 100% consistent in
choosing between 1) and 2), but have tried to rigorously avoid 3).  In the specific case of
the t-test, I think I erred in setting the minimum sample size to 5, since the test can in
fact be meaningfully performed (though power will be very low) for samples as small as 2.
 I will fix this.

>2) I cannot find the functions for a homoscedastic or paired t-test (I
>assumed the one that is implemented is for an unequal variance). There are
>3 kinds of T-Tests and it is not clear to see which one is used. I would
>love to see something like double ttest(double[] v1, double[] v2, int type)
>where type is TTest.PAIRED or TTest.HOMOSCEDASTIC or >TTest.HETEROSCEDASTIC

Good point.  The reference in the javadoc gives the formulas used, which are for un-paired,
no assumption of equal variance.  I will make this more explicit in the javadoc itself.  Regarding
the paired and homoscedastic tests, I agree that this would be a good addition and it seems
reasonable to me to add a flag to indicate the type of the test.  Patches are welcome :-)
 If there are no objections, and no patches come in this week, I will try to ge this in this
weekend.  

Phil




_________________________
Joel Freyss
*************************************************************************
The information of this e-mail and in any file transmitted with it is
strictly confidential and may be legally privileged.
It is intended solely for the addressee. If you are not the intended
recipient, any copying, distribution or any other use of this e-mail is
prohibited and may be unlawful. In such case, you should please notify the
sender immediately and destroy this e-mail.
The content of this e-mail is not legally binding unless confirmed by
letter. Any views expressed in this message are those of the individual
sender, except where the message states otherwise and the sender is
authorised to state them to be the views of the sender's company.
 For further information about Actelion please see our website at
http://www.actelion.com.
**************************************************************************


---------------------------------------------------------------------
To unsubscribe, e-mail: commons-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-dev-help@jakarta.apache.org



Mime
View raw message