Hello to everyone,
I found some interesting information on the net.
The *National Institute of Standards* (NIST) publishes some *test cases*
for statistical software to test the reliability of the statistical
algorithms implemented. You can find more information at
http://www.itl.nist.gov/div898/strd/ ; further information can be found
at the following address, too:
http://www.csdassn.org/software_reports/gnumeric.pdf
To quote the NIST:
The purpose of this project is to improve the accuracy of
statistical software
by providing reference datasets with certified computational
results that enable
the objective evaluation of statistical software.
USED the NumAcc4 TESTSUITE: (see
http://www.itl.nist.gov/div898/strd/univ/certvalues/numacc4.html )
OK, I have tried the most difficult test case, AND Calc fails (for
CORREL major failure):
- for the mean at the 7th decimal digit
- for SD: at the 10th decimal digit
- I believe Calc should do a little better (at least 12th decimal digit
for the mean)
- accepted results for professional software are usually 14-15 decimal
digits
The correlation coefficient fails poorly:
- made column B = A+1
- C1: =CORREL(A1:A1001; B1:B1001)
- reported correl coeff. = -0.882
- should have been = -0.999
When I will have some more spare time, I will do the remaining tests, too.
Kind regards,
Leonard Mada
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
