[S] non-linear regression with user-defined loss function

[ArbuthnotC <ArbuthnotC@aol.com>]

I would like to carry out a non-linear regression analysis using a user-
defined loss function:i.e. instead of least squares:

loss=(obs-pred)^2

I want to minimise an weighted expresssion such as:

loss=1000*(obs-pred)^2            if obs<pred
loss=(obs-pred)^2                    if obs=>pred

Any ideas??

Thanks

colin arbuthnot  

 
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu.  To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message:  unsubscribe s-news