[Jprogramming] Stats example ...

['Bo Jacoby' via Programming]

   E+/ % # E 5 2 6 75 (-E) 5 2 6 70 _3 1 2 *:(-E) 5 2 6 70 9 1 4 E*:(-E) 5 2 6 73.5 E&.:*:(-E) 5 2 6 71.87083
 

    Den 18:33 mandag den 13. juni 2016 skrev Martin Kreuzer <info at airkreuzer.com>:
 
 

 Hi all -

While browsing the J NuVoc page I accidentally stumbled upon this example:

URL: http://code.jsoftware.com/wiki/Vocabulary/ampdotco

Under "Common Uses / 2. Compute standard deviation as the 
root-mean-square of a list of differences from the mean value"
the example starts out  with

y =: 0 _2 1 2          NB. list of differences from the mean value

This didn't look right to me as the sum of deviations does not result in zero:

    +/ 0 _2 1 2
1

So I made up an example myself:

    obs=. 5 2 6 7    NB. observed data
    amean=. +/ % #    NB. arithmetic mean
    dev=. ] - amean    NB. deviation from mean value
    dev obs
0 _3 1 2

Comparing this result with the above (original set of deviations)
makes me think it might be a typo.

The "mean deviation" (mean of absolute deviation values) would then be

    dmean=. amean @: |
    dmean dev obs
1.5

and the standard deviation (root-mean-square of deviation values)

    stddev=. amean &.: *:
    stddev dev obs
1.87083

Would somebody please have a look into that to check whether I'm 
utterly wrong or not..?

Thanks
-M
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm