An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20110711/f155e5bc/attachment.pl>
problem finding p-value for entropy in reldist package
3 messages · Amy Wesolowski, VictorDelgado, Uwe Ligges
Hi Amy Wesolowski, I don't have a straightfoward answer to you question. I have been working with reldist too, and the 'rpy' and 'rpluy' functions described by "Applying Relative Distribution Methods in R" are also not working here in my 2.9.1 R-version. I think its because they are reldist internall function, so, maybe its possible that they only work with previous objects and set ups... But if you look to internal parametres of reldist, you could set "ci = TRUE", it's constructs the confidence interval for entropy by the proportion of original cohort. It's still unhelpfull to understand how this intervall is constructed, and also does not show the overall interval, but you can se the values with $ci. By the Handcock and Morris (1998) paper is posible to intuit that they are comparing the 0.00 entropy with the 95% Confidence Interval around the estimate. For example, in this artigle, pag. 74, they reach an overall entropy of 0.125, the lower 95%_CI is 0.092. The 0.00 comparision is far below this lower bound, so its resonable to think the p-value is realy 0.000. But it's only a clue to approximate the true p-value. But we still needing to see: 1) how this intervall is constructed (I have no idea what distribution the entropy should have, and if it changes by data) and 2) Knowing the first point, how to set alpha values). Good luck, Victor Delgado cedeplar.ufmg.br P.H.D. student www.fjp.mg.gov.br reseacher -- View this message in context: http://r.789695.n4.nabble.com/problem-finding-p-value-for-entropy-in-reldist-package-tp3659806p3660228.html Sent from the R help mailing list archive at Nabble.com.
Have you noted you sent your message to the R-help list only and forgot to include the original poster? You also forgot to cite the original question (and any other former parts of the thread as far as there was any). Please do so when sending messages to a mailing list such as R-help. Thanks, Uwe Ligges
On 11.07.2011 19:24, VictorDelgado wrote:
Hi Amy Wesolowski, I don't have a straightfoward answer to you question. I have been working with reldist too, and the 'rpy' and 'rpluy' functions described by "Applying Relative Distribution Methods in R" are also not working here in my 2.9.1 R-version. I think its because they are reldist internall function, so, maybe its possible that they only work with previous objects and set ups... But if you look to internal parametres of reldist, you could set "ci = TRUE", it's constructs the confidence interval for entropy by the proportion of original cohort. It's still unhelpfull to understand how this intervall is constructed, and also does not show the overall interval, but you can se the values with $ci. By the Handcock and Morris (1998) paper is posible to intuit that they are comparing the 0.00 entropy with the 95% Confidence Interval around the estimate. For example, in this artigle, pag. 74, they reach an overall entropy of 0.125, the lower 95%_CI is 0.092. The 0.00 comparision is far below this lower bound, so its resonable to think the p-value is realy 0.000. But it's only a clue to approximate the true p-value. But we still needing to see: 1) how this intervall is constructed (I have no idea what distribution the entropy should have, and if it changes by data) and 2) Knowing the first point, how to set alpha values). Good luck, Victor Delgado cedeplar.ufmg.br P.H.D. student www.fjp.mg.gov.br reseacher -- View this message in context: http://r.789695.n4.nabble.com/problem-finding-p-value-for-entropy-in-reldist-package-tp3659806p3660228.html Sent from the R help mailing list archive at Nabble.com.
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.