[R] lmom package - Resending the email
Katherine Gobin
katherine_gobin at yahoo.com
Fri Dec 5 06:10:41 CET 2014
Dear Dalgaard sir,
Thanks a lot for detailed clarification. It indeed is very enlightening and will be very useful for me in future.
And your suggestion is well taken.
Thanks again.
Regards
Katherine
--------------------------------------------
On Thu, 4/12/14, peter dalgaard <pdalgd at gmail.com> wrote:
Subject: Re: [R] lmom package - Resending the email
To: "Simon Zehnder" <szehnder at uni-bonn.de>
<r-help at r-project.org>
Date: Thursday, 4 December, 2014, 2:04 PM
lmom is based on
L-moments, which are different from ordinary moments, except
for the 1st one. It would be truly miraculous if it gave the
same result as the ordinary method of moments or maximum
likelihood.
Estimates of
any distributional parameter requires that the model
actually fits the data, and in your case a qqnorm(amounts)
shows that they are certainly not normal. In such cases, the
L-moment estimator of the std.dev. is not necessarily an
estimate of the std.dev. of the actual distribution.
A lognormal distribution seems
to fit the data better. However, the L-moments suggest a
value for zeta (the lower bound) of 3226 which is well
inside the range of the actual data. In fact there are 16
observations that are less than 3226. Maximum likelihood
would never do that, but the same sort of effect is
well-known for the ordinary method of moments.
In short, you need to study
the theory before you appply its results.
- Peter D.
On 03 Dec 2014, at 10:57 ,
Simon Zehnder <szehnder at uni-bonn.de>
wrote:
> Katherine,
>
> for a deeper
understanding of differing values it makes sense to provide
the list at least with an online description of the
corresponding functions used in Minitab and SPSS…
>
> Best
> Simon
> On 03 Dec 2014,
at 10:45, Katherine Gobin via R-help <r-help at r-project.org>
wrote:
>
>> Dear R
forum
>> I sincerely apologize as my
earlier mail with the captioned subject, since all the
values got mixed up and the email is not readable. I am
trying to write it again.
>> My
problem is I have a set of data and I am trying to fit some
distributions to it. As a part of this exercise, I need to
find out the parameter values of various distributions e.g.
Normal distribution, Log normal distribution etc. I am using
lmom package to do the same, however the parameter values
obtained using lmom pacakge differ to a large extent from
the parameter values obtained using say MINITAB and SPSS as
given below -
>>
_____________________________________________
>>
>> amounts =
c(38572.5599129508,11426.6705314315,21974.1571641187,118530.32782443,3735.43055996748,66309.5211176106,72039.2934132668,21934.8841708626,78564.9136114375,1703.65825161293,2116.89180930203,11003.495671332,19486.3296339113,1871.35861218795,6887.53851253407,148900.978055447,7078.56497101651,79348.1239806592,20157.6241066905,1259.99802108593,3934.45912233674,3297.69946631591,56221.1154121067,13322.0705174134,45110.2498756567,31910.3686613912,3196.71168501252,32843.0140437202,14615.1499458453,13013.9915051561,116104.176753387,7229.03056392023,9833.37962177814,2882.63239493673,165457.372543821,41114.066453219,47188.1677766245,25708.5883755617,82703.7378298092,8845.04197017415,844.28834047836,35410.8486123933,19446.3808445684,17662.2398792892,11882.8497070776,4277181.17817307,30239.0371267968,45165.7512343364,22102.8513746687,5988.69296597127,51345.0146170238,1275658.35495898,15260.4892854214,8861.76578480635,37647.1638704867,4979.53544046949,7012.48134772332
,3385.20612391205,1911.03114395959,66886.5036605189,2223.47536156462,814.947809578378,234.028589468841,5397.4347625133,13346.3226579065,28809.3901352898,6387.69226236731,5639.42730553242,2011100.92675507,4150.63707173462,34098.7514446498,3437.10672573502,289710.315303182,8664.66947305203,13813.3867161134,208817.521491857,169317.624400274,9966.78447705792,37811.1721605562,2263.19211279927,80434.5581206454,19057.8093104899,24664.5067589624,25136.5042354789,3582.85741610706,6683.13898432794,65423.9991390846,134848.302304064,3018.55371579808,546249.641168158,172926.689143006,3074.15064180208,1521.70624812788,59012.4248281661,21226.928522236,17572.5682970983,226.646947337851,56232.2982652019,14641.0043361533,6997.94414914865)
>>
>>
library(lmom)
>> lmom =
samlmu(amounts)
>> #
__________________________________________________________________
>> # Normal Distribution parameters
>> parameters_of_NOR <-
pelnor(lmom); parameters_of_NOR
>>
>> mu sigma
115148.4 175945.8
>>
Location Scale
Minitab 115148.4
485173SPSS 115148.4
485173
>> #
__________________________________________________________________
>> # Log Normal (3 Parameter)
Distribution parameters
>>
zeta mu
sigma 3225.798890 9.114879
2.240841
>>
Location Scale
Shape
>> MINITAB
9.73361
1.76298 75.51864SPSS
9.7336 1.763
75.519 #
__________________________________________________________________
>>
>> Besides
Genaralized extreme Value distributions, all the other
distributions e.g. Gamma, Exponential (2 parameter)
distributions etc give different results than MINITAB and
SPSS.
>> Can some one guide me?
>>
>> Regards
>> Katherine
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> [[alternative HTML version
deleted]]
>>
>>
______________________________________________
>> R-help at r-project.org
mailing list -- To UNSUBSCRIBE and more, see
>> 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.
>
>
______________________________________________
> R-help at r-project.org
mailing list -- To UNSUBSCRIBE and more, see
> 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.
--
Peter Dalgaard,
Professor,
Center for Statistics, Copenhagen
Business School
Solbjerg Plads 3, 2000
Frederiksberg, Denmark
Phone:
(+45)38153501
Email: pd.mes at cbs.dk Priv:
PDalgd at gmail.com
More information about the R-help
mailing list