- PARMS <name_list [=numbers]
[, name_list [=numbers] ...
</ options> ;
The PARMS statement lists names of parameters and specifies initial
values, possibly over a grid. You can specify the parameters and
values either directly in a list or provide the name of a SAS data set
that contains them using the DATA= option.
While the PARMS statement is not required, you are encouraged to
use it to provide PROC NLMIXED with accurate starting values.
Parameters not listed in the PARMS statement are assigned an
initial value of 1. PROC NLMIXED considers all symbols
not assigned values to be parameters, so you should specify
your modeling statements carefully and check the output from
the "Parameters" table to make sure the proper parameters are
A list of parameter names in the PARMS statement is not separated by
commas and is followed by an equal sign and a list of numbers. If
the number list consists of only one number, this number defines the
initial value for all the parameters listed to the left of the equal
If the number list consists of more than one number, these numbers
specify the grid locations for each of the parameters listed to the
left of the equal sign. You can use the TO and BY keywords to
specify a number list for a grid search. If you specify a grid of
points in a PARMS statement, PROC NLMIXED computes the objective
function value at each grid point and chooses the best (feasible)
grid point as an initial point for the optimization process. You
can use the BEST= option to save memory for the storing and sorting
of all grid point information.
The following options are available in the PARMS statement after
a slash (/):
specifies the maximum number of points displayed in the
"Parameters" table, selected as the points with the maximum
likelihood values. By default, all grid values are displayed.
specifies a SAS data set
containing parameter names and starting values. The data set should
be in one of two forms: narrow or wide. The narrow-form data set
contains the variables PARAMETER and ESTIMATE, with parameters and
values listed as distinct observations. The wide-form data set has
the parameters themselves as variables, and each observation provides
a different set of starting values. BY groups are ignored in this
data set, so the same starting grid is evaluated for each BY group.
Copyright © 1999 by SAS Institute Inc., Cary, NC, USA. All rights reserved.