[
https://issues.apache.org/jira/browse/MATH-1401?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16820531#comment-16820531
]
Michele De Stefano commented on MATH-1401:
------------------------------------------
Dear Gilles,
I've just attached the modified unit test (ClopperPearsonIntervalTest.java) and
the modified (now working) code (ClopperPearsonInterval.java).
Your fix was failing to address the case with number of successes = 0.
Because I am also a data scientist and I have a lot of experience in Python
scientific programming, I generated 3 meaningful test cases with expected
results obtained by the same algorithm implemented in a Python package that I
use (the package name is statsmodels). I use that package since a while and I
trust its results (for our specific case, I've also partially verified them
theoretically).
So, in the unit test you will find comments from myself, explaining which was
the Python instruction I used to produce the expected numbers (so that anyone
that is also able to program in Python can reproduce them).
I suspect (but I have not verified this) that we have similar issues with the
implementation of the other algorithms that produce confidence intervals for
the binomial distribution.
In summary, the 3 cases we need to test for each of these methods are
# number of successes = 0, number of trials = N
# number of successes = k, number of trials = N, 0 < k < N
# number of successes = number of trials = N
The code I've attached now passes all the three of them.
Best regards,
Michele
[^ClopperPearsonIntervalTest.java]
[^ClopperPearsonInterval.java]
> Exception at IntervalUtils.getClopperPearsonInterval
> ----------------------------------------------------
>
> Key: MATH-1401
> URL: https://issues.apache.org/jira/browse/MATH-1401
> Project: Commons Math
> Issue Type: Bug
> Affects Versions: 3.6.1
> Reporter: Art
> Assignee: Bruno P. Kinoshita
> Priority: Major
> Fix For: 4.0
>
> Attachments: ClopperPearsonInterval.java,
> ClopperPearsonIntervalTest.java
>
>
> IntervalUtils.getClopperPearsonInterval throws an exception when number of
> successes equals to zero or number of successes = number of trials.
> IntervalUtils.getClopperPearsonInterval(1, 0, 0.95) or
> IntervalUtils.getClopperPearsonInterval(1, 1, 0.95) throws
> org.apache.commons.math3.exception.NotStrictlyPositiveException despite that
> its input parameters are valid.
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
