[R] e1071::skewness and psych::skew return NaN

[Rui Barradas]

Hello,

Just remembered, you can see how many values are zero, and since it's 
only one value, remove it and log the rest.

sum(dat == 0) # 1
d2 <- dat[dat != 0]

library(psych)
skew(log(d2))
[1] 0.6089985


Hope this helps,

Rui Barradas


Em 13-02-2013 17:59, Rui Barradas escreveu:
> Hello,
>
> That value means that some values of your data are negative or zero. A
> simple inspection shows that
>
> any(dat < 0)  # FALSE
> any(dat == 0) # TRUE
>
> Solution: don't log your data
>
>
> Hope this helps,
>
> Rui Barradas
>
> Em 13-02-2013 16:55, Stephen Politzer-Ahles escreveu:
>> Hello everyone,
>>
>> Does anyone know what would cause the skewness() function (from
>> e1071), as well as skew() from psych, to return a value of NaN?
>>
>> I have a vector of positively-skewed data
>> (https://docs.google.com/file/d/0B6-m45Jvl3ZmYzlHRVRHRURzbVk/edit?usp=sharing)
>>
>> which these functions return a value for like normal:
>>
>>> skewness( data ) # returns 1.400405
>>
>> but when I instead give those functions the log-transformed data they
>> return NaN
>>
>>> skewness( log( data ) ) #returns NaN
>>
>> The same occurs when I feed the function data transformed by reflected
>> reciprocal
>>
>>> skewness( max(data) - 1/data ) ) #returns NaN
>>
>> The vector has no missing values (and if it did, I would get NA rather
>> than NaN, and the function wouldn't return a number when I give it the
>> raw data).
>>
>> Best,
>> Steve
>>
>> --
>> Stephen Politzer-Ahles
>> University of Kansas
>> Linguistics Department
>> http://people.ku.edu/~sjpa/
>>
>> ______________________________________________
>> 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.
>>