On Mon, Jul 9, 2012 at 9:58 AM, Chris Marshall <[email protected]>wrote:
> On Fri, Jul 6, 2012 at 12:31 PM, David Mertens <[email protected]> > wrote: > > ...snip... > > > > That is the so-called population standard deviation, which assumes that > you > > got the average value from the population, not the sample. For large data > > sets, the discrepancy is not important. For small data sets, this is only > > accurate if you have the entire population of interest. If you only have > a > > subset of the population, then the statistically correct value comes by > > dividing instead by the number of elements minus 1, i.e. > > > > my $st_dev = sqrt( $sq_diff_sum / ($sq_diff_sum->nelem - 1) ); > > > > In this case---finding the mean and standard deviation of the elements > along > > the diagonal---either method is acceptable, and you should report which > > value you calculated when you show your results. For sample-based > > statistical calculations, where you only take a sample of your whole > > population, the latter calculation is correct. > > > > I say all of this just to point out that the second return value of the > > stats and statsover methods is the sample RMS deviation from the mean, > i.e. > > it divides by N-1. (The docs call this the population RMS deviation for > > reasons not clear to me.) If you want to use the population RMS, you > should > > get the 7th argument, like so: > > This seems like a documentation problem to me. > In fact, I would prefer that the RMS be the second > value and have the sample RMS one at the end. > > --Chris > I agree. But how do we move forward? One way is to create new functions called "stat" and "statover" that do what we mean, and deprecate "stats" and "statsover" Thoughts? David -- "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it." -- Brian Kernighan
_______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
