Re: linear regression statistics

There is no built-in function to calculate the 'cod'[R^2]

However, you could write your own.

Something like:

Total_Variance = variance( z ) ; z[*] is the original series

If 'Z'represents the estimates

    rc = regline (x,z)

    Residual_Variance = (z - (rc*x + rc_at_yintercept)

    cod =1.-(Residual_Variance/Total_Variance)

Good luck

Jesse Miller wrote:
> Is there an NCL function to help calculate the coefficient of
> determination (R-squared) between observed data and a linear regression
> plot calculated via regline?
> Thanks.
> Jesse
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Tue Oct 06 2009 - 10:23:44 MDT