Wrong formula in CORR documentation

lukaseder
SQL-Fighter

The CORR documentation lists a formula defining

 

 

 

CORR(x, y) = COVAR_POP(x, y) / (STDDEV_POP(x) * STDDEV_SAMP(y))

 

 

 

When in fact, it should be:

 

 

 

CORR(x, y) = COVAR_POP(x, y) / (STDDEV_POP(x) * STDDEV_POP(y))

 

 

 

See: https://docs.exasol.com/sql_references/functions/alphabeticallistfunctions/corr.htm

While it is possible to calculate the coefficient for a sample rather than a population, I don't think the two types of functions should be mixed, see also: https://en.wikipedia.org/wiki/Pearson_correlation_coefficient 

1 ACCEPTED SOLUTION

Accepted Solutions

exa-SimonR
Team Exasol
Team Exasol

Hi Lukas,

thanks again for the bug reports. You're right that the documentation is incorrect. The implementation uses the correct formula.

I will create an internal ticket for our documentation team to fix the issue.

Cheers,

Simon

View solution in original post

1 REPLY 1

exa-SimonR
Team Exasol
Team Exasol

Hi Lukas,

thanks again for the bug reports. You're right that the documentation is incorrect. The implementation uses the correct formula.

I will create an internal ticket for our documentation team to fix the issue.

Cheers,

Simon

View solution in original post