[pymvpa] normalise

[Nick Oosterhof]

On Aug 20, 2014, at 12:29 PM, cfygj <cfygj at 126.com> wrote:

> I have used the command line like zscore(ds, chunks_attr='chunks', param_est = ('targets',['0'])).

Note on terminology: by "the command line", most people mean a means to interact with the computer by issuing commands in successive lines of text.
What you refer to as "the command line" (the 'zscore...' string would be just a "command").

> However, the elements of data
> were replaced with [[Naf, Naf, Naf,...],[Naf, Naf,...],[Naf, Naf, Naf,...]...]. So I must modify this command line. But I don't know how do it.

the zscore function, when provided with a param_est parameter, uses all samples that match the param_est value to estimate the mean and standard deviation.
Probably in your case there are no targets with value '0', hence these values cannot be estimated.

If you want to use /all/ samples to estimate the mean and standard deviation, then run the command without param_est.

You could have read all of this in the documentation. I googled for 'pymvpa zscore' and the first hit was http://www.pymvpa.org/generated/mvpa2.mappers.zscore.zscore.html