Hello What is the appropriate method to test if a given distribution is a weibull thank you meriema email meriema.belaidouni at int-evry.fr
weibull test
3 messages · Meriema Belaidouni, Emmanuel Paradis, Kjetil Halvorsen
At 11:18 18/12/2002 +0100, vous avez ?crit:
Hello What is the appropriate method to test if a given distribution is a weibull thank you meriema email meriema.belaidouni at int-evry.fr
You can use a Kolmogorov-Smirnov test. See ?ks.test where there is an example with a gamma distribution. You will probably need to estimate the parameters of the Weibull which you can do with fitdistr() in package MASS. However, I don't know if the distribution of the KS test will be the same if some parameters are estimated from the data or not (apparently, ks.test() doesn't do any correction). Someone on the list can help here. Emmanuel Paradis
I don't know the answer, but there must be something better than the KS test?, but maybe not in R. If you can think about an alternative you can construct a likelihood ratio test. Kjetil Halvorsen
Emmanuel Paradis wrote:
At 11:18 18/12/2002 +0100, vous avez ?crit:
Hello What is the appropriate method to test if a given distribution is a weibull thank you meriema email meriema.belaidouni at int-evry.fr
You can use a Kolmogorov-Smirnov test. See ?ks.test where there is an example with a gamma distribution. You will probably need to estimate the parameters of the Weibull which you can do with fitdistr() in package MASS. However, I don't know if the distribution of the KS test will be the same if some parameters are estimated from the data or not (apparently, ks.test() doesn't do any correction). Someone on the list can help here. Emmanuel Paradis
______________________________________________ R-help at stat.math.ethz.ch mailing list http://www.stat.math.ethz.ch/mailman/listinfo/r-help