[R] Reversing data transformation

Bert Gunter gunter.berton at gene.com
Sat Apr 6 15:46:26 CEST 2013


Sorry --failed to cc the list. -- Bert

On Fri, Apr 5, 2013 at 9:29 PM, Bert Gunter <bgunter at gene.com> wrote:

 Have you tried plotting your data?!

>library(lattice)
>densityplot(yourdata)

 Forget tests of normality -- they are a BAD idea, despite what the
 statistics textbooks say (IMHO obviously) -- and focus on what the
 data are trying to tell you, which sure ain't that they are a sample
 from a normal distribution! Subject matter knowledge and problem
 context -- and probably explicit knowledge of how the data were
 produced -- are probably essential.

 -- Bert

>
> On Fri, Apr 5, 2013 at 3:47 PM, Beatriz González Domínguez
> <aguitatierra at hotmail.com> wrote:
>> Hi everybody,
>>
>> I would be very grateful if you could give me your thoughts on the following issue.
>>
>> I need to perform  Box-Cox (bcPower€ ) transformation on my data. To do this, I calculated lambda using the function '€ powerTransform'€ .
>>   powerTransform(data)
>> However, I got an error message when performing this function:
>>   Convergence failure: return code = 52
>> I was told by John Fox that the nature of my data was probably causing the problem so I applied the following transformation:
>>   powerTransform(1e7*(data - 0.99*(min(data))))   [1]
>> And it worked! However, now when I try to check the normality of my data,
>>   shapiro.test(data)
>> I get a new warning message:
>>   Error in shapiro.test(data) :
>>     ifault=6. This should not happen
>> I think this may be solved by removing the transformation I applied previously in [1], but I’m not sure how to do it properly.
>>
>> Could anyone give me some advice?
>>
>> MY DATA:
>  c(0.000103113, 0.000102948, 0.000104001, 0.000103794, 0.000104628,
>  9.2765e-05, 9.4296e-05, 9.5025e-05, 9.4978e-05, 9.8821e-05, 9.7586e-05,
>  9.6285e-05, 0.00010158, 0.000100919, 0.000103535, 0.000103321,
>  0.000102842, 0.000102315, 0.0001033, 0.000103691, 0.000102689,
>  0.000103248, 0.000101216, 0.000103127, 0.000103439, 0.000102818,
>  0.000103482, 0.00010227, 0.000104264, 0.000104672, 0.000104586,
>  0.000104516, 0.000104585, 0.000103804, 0.000103702, 0.000103405,
>  0.000104815, 0.000101718, 0.000101792, 0.000103273, 0.000102942,
>  0.000102522, 0.000103149, 0.0001036995, 0.0001039735, 0.000101487666666667,
>  0.000102796)
>>
>> Thanks a lot!!
>>
>> Bea
>>
>>
>>
>>
>>         [[alternative HTML version deleted]]
>>
>>
>> ______________________________________________
>> 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.
>>
>
>
>
> --
>
> Bert Gunter
> Genentech Nonclinical Biostatistics
>
> Internal Contact Info:
> Phone: 467-7374
> Website:
> http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm



-- 

Bert Gunter
Genentech Nonclinical Biostatistics

Internal Contact Info:
Phone: 467-7374
Website:
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm



More information about the R-help mailing list