# Model functions

## units

units
creates a factor with a level for every record in the data file.
This is used to fit the 'nugget' variance when a correlation structure is applied to the residual.
## uni() function

uni(`f,k`[,`n`])
creates a factor with a level for every record
subject to the factor level of `f`
equalling `k`, i.e. a new level
is created for the factor whenever a new record is encountered whose
integer truncated data value from data field `f`
is `k`.
Thus uni(site,2) would be used to create an independent
error term for site 2 in a multi-environment trial
and is equivalent to at(site,2).units. The default size
of this model term is the number of data records. The user may
specify a lower number as the third argument. There is little
computational penalty from the default but the .sln file
may be substantially larger than needed.
uni(`f`[,0[,`n`]])
creates a factor with a new level whenever there is a level
present
for the factor `f`.
Levels (effects) are not
created if the level of factor `f` is 0, missing
or negative.
The size may be set in the third
argument by setting the second argument to zero.
**Return to start**