The NNET Procedure
PROC NNET Statement
PROC NNET <options>;
The PROC NNET statement invokes the procedure. You can specify the following options in the PROC NNET statement:
You can specify the following options:
- DATA=CAS-libref.data-table
names the input data table for PROC NNET to use. The default is the most recently created data table. CAS-libref.data-table is a two-level name, where
- CAS-libref
refers to a collection of information that is defined in the LIBNAME statement and includes the
caslib, which includes a path to the data, and a session identifier, which defaults to the active session but which can be explicitly defined in the LIBNAME statement. For more information about CAS-libref, see the section Using CAS Sessions and CAS Engine Librefs.- data-table
specifies the name of the input data table.
Note: The data set options WHERE, TEMPNAMES, and TEMPEXPRESS are not supported.
- INMODEL=CAS-libref.data-table
specifies a model for stand-alone scoring or coding. CAS-libref.data-table is a two-level name, where CAS-libref refers to the
casliband session identifier, and data-table specifies the name of the input data table. For more information about this two-level name, see the DATA= option and the section Using CAS Sessions and CAS Engine Librefs.- LISTNODE=INPUT |HIDDEN |OUTPUT
specifies the values of the layers to be saved in the scored output or code. Values are normalized values after standardization, if any. This option is valid only if you specify the OUTPUT or CODE statement.
You can specify the following values:
- INPUT
outputs the values of the input layer.
- HIDDEN
outputs the values of the hidden layer(s).
- OUTPUT
outputs the values of the output layer.
By default, LISTNODE=HIDDEN for autoencoder networks. For non-autoencoder networkers, no layer information is saved if the LISTNODE= option is not specified.
- MISSING=MAX |MEAN |MIN
specifies the statistic with which to impute missing values for interval input variables. If you specify this option, missing values for nominal input variables are treated as a valid level. PROC NNET excludes any observations that have missing values in the target variable.
You can specify the following values:
- MAX
substitutes missing values with the maximum value of the corresponding interval input variable.
- MEAN
substitutes missing values with the mean value of the corresponding interval input variable.
- MIN
substitutes missing values with the minimum value of the corresponding interval input variable.
By default, observations that have missing values are excluded from the analysis.
- NTHREADS=number-of-threads
specifies the number of threads to use for the computation. The default value is the number of CPUs available. The value of number-of-threads can be from 1 to 64, inclusive.
- PRINTTARGET
prints out the ODS tables that contain the mapping of the target variable or its events to variables PROC NNET predicts. The ODS names are PredProbName, PredIntoName, and PredName in Table 3. This option is valid when TARGET statement is specified with OUTPUT statement or CODE statement.
- STANDARDIZE=NONE |STD |MIDRANGE
specifies the method to use for standardizing the interval inputs.
You can specify the following values:
- NONE
does not alter the variables.
- STD
scales the variables such that their mean is 0 and the standard deviation is 1.
- MIDRANGE
scales the variable such that their midrange is 0 and the half-range is 1. That is, the variables have a minimum of –1 and a maximum of 1. You can also use the keyword RANGE for this method.
By default, STANDARDIZE=MIDRANGE.
-
TARGETSTANDARDIZE=NONE |STD |MIDRANGE
TARGETSTD=NONE |STD |MIDRANGE specifies the method to use for standardizing the interval target.
You can specify the following values:
- NONE
does not alter the variables.
- STD
scales the variables such that their mean is 0 and the standard deviation is 1.
- MIDRANGE
scales the variable such that their midrange is 0 and the half-range is 1. That is, the variables have a minimum of –1 and a maximum of 1. You can also use the keyword RANGE for this method.
By default, TARGETSTANDARDIZE=MIDRANGE.