ASSESS Procedure
PROC ASSESS Statement
PROC ASSESS <options>;
The PROC ASSESS statement invokes the ASSESS procedure.
You can specify the following options:
- DATA=libref.data-table
-
names the input data table for PROC ASSESS to use. The default is the most recently created data table. libref.data-table is a two-level name, where
- libref
refers to a collection of information that is defined in the LIBNAME statement and includes the
library, 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 libref, see the section Using CAS Sessions and CAS Engine Librefs.- data-table
specifies the name of the input data table.
- FITSTATOUT=libref.data-table
-
specifies the name of the fit statistics results table. libref.data-table is a two-level name, where libref refers to the library, and data-table specifies the name of the output data table. For more information about this two-level name, see the DATA= option and the section Using CAS Sessions and CAS Engine Librefs.
If you specify this option, all ODS tables are suppressed. This option is valid only if you also include a FITSTAT statement.
- LIFTOUT=libref.data-table
-
specifies the name of the lift results table. libref.data-table is a two-level name, where libref refers to the library, and data-table specifies the name of the output data table. For more information about this two-level name, see the DATA= option and the section Using CAS Sessions and CAS Engine Librefs.
If you specify this option, all ODS tables are suppressed.
- MAXITER=integer
specifies the maximum number of iterations for the percentile algorithm, where integer must be an integer greater than or equal to 1. By default, MAXITER=5 times the value of the NBINS= option.
- METHOD=EXACT |ITERATIVE
-
specifies the analytic method to use to compute the percentile for binning in the lift analysis.
You can specify the following values:
- EXACT
uses smart technology to perform an exact count to compute the percentile without global sorting in the distributed environment.
- ITERATIVE
uses the iterative approach to approximate the percentile in the distributed environment.
If you specify METHOD=EXACT, the MAXITER= option is ignored. By default, METHOD=ITERATIVE.
- NBINS=integer
-
specifies the number of bins to use in the lift calculation, where integer must be an integer greater than or equal to 2.
By default, NBINS=20.
- NCUTS=integer
-
specifies the number of cuts to use in the ROC calculation, where integer must be an integer greater than or equal to 2. The inverse of integer represents the step size of the quantile. For example, NCUTS=10 generates 10 intervals for ROC analysis, with each step size 0.1. You cannot specify this option when the response variable that is specified in the TARGET statement is interval.
By default, NCUTS=100.
- NOMISSINGTARGET
excludes the observations that have missing values of a nominal target from assessment analysis.
- NTHREADS=number
specifies the number of threads to use in the computation, where number must be an integer between 1 and 64, inclusive. The default value is the number of CPUs available in the machine.
- QUANTILEBIN
improves binning allocations in the lift analysis in certain situations where data are not evenly distributed, with spikes in some intervals.
- ROCOUT=libref.data-table
-
specifies the name of the ROC results table. libref.data-table is a two-level name, where libref refers to the library, and data-table specifies the name of the output data table. For more information about this two-level name, see the DATA= option and the section Using CAS Sessions and CAS Engine Librefs.
If you specify this option, all ODS tables are suppressed. This option is valid only for a nominal target (LEVEL=NOMINAL option in the TARGET statement).
- USERCUTOFF=ratio
-
specifies a user-defined threshold for events, where ratio must be between 0 and 1, exclusive.
If you specify this option, the ROC table includes only one row, which contains the user-defined cutoff value, and the NCUTS= option is ignored.
- USEDATARANGE(NROCBINS=number LB=number UB=number)
-
computes the ROC-related statistics on the basis of the specified range of the input data.
You can specify the following additional suboptions:
- NROCBINS=number
-
specifies the number of bins to apply when you compute ROC-related statistics on the basis of the range of the input data.
By default, NROCBINS=5000.
- LB=number
-
specifies the lower-bound value of the range of the input data to apply when you compute ROC-related statistics, where number must be between 0 and 1, inclusive.
By default, LB=0.
- UB=number
-
specifies the upper-bound value of the range of the input data to apply when you compute ROC-related statistics, where number must be between 0 and 1, inclusive.
By default, UB=1.
If neither the LB= option nor UB= option is specified, the ROC-related statistics are computed on the basis of the range of the input data.