The DMREG Procedure
SCORE Statement
Specifies options for scoring data.
SCORE scoring-option(s);
Options
Scoring-options can be the following:
ADDITIONALRESIDUALS
Specifies that the OUT= data set contains additional residuals such as: RS_ for logistic regressions
and RS_, RT_, RD_, RDS_, RDT_ for normal error regression. See for more detail.
Alias:
ADDRES |AR
ALPHA=number
Specifies the significance level p for the construction of 100(1-p)% confidence interval for the
posterior probabilities. This number must be between 0 and 1.
Default:
0.05
CLP
Specifies that the OUT= data set contains the confidence limites for the posterior probabilities.
The significance leve is controlled by the ALPHA= option.
DATA= SAS-data-set
Specifies the input data set that contains inputs and optionally targets.
Default:
The default is the same as the DATA= data set in the PROC statement.
DMDB | NODMDB
Specifies whether an explicit DATA= data set has been DMDB-encoded or if the data set contains
raw data.
Default:
If the DATA= option is not specified in the SCORE statement, the training
data is used and the NODMDB option is invalid.
Caution:
If the DATA= in the SCORE statement specifies a data set other than the
training data, either DMDB or NODMDB must be specified in the SCORE
statement.
OUT= SAS-data-set
Specifies the output data set with outputs.
Default:
DATAn
Names for computed variables are normally taken from the data dictionary. If necessary, names
for these variables can be generated by concatenating a prefix to the name of the corresponding
target variable according to the rules in the following tables:
Statistics Generated in the
OUT=SAS-data-set for Normal
EerrorRegression
NAME
LABEL
P_targetname
Predicted:
targetname
E_targetname
Error
Function:
targetname
R_targetname
Residuals:
targetname
RD_targetname
Deviance
Residuals:
targetname
If the target is declared as a
categorical variable, the
OUT=SAS-data-set also includes:
F_targetname
From:
targetname
I_targetname
Into:
targetname
If the ADDITIONALRESIDUALS
option is also specified, the
OUT=SAS-data-set includes:
RS_targetname
Standardized
Residuals:
targetname
RT_targetname
Studentized
Residuals:
targetname
RDS_targetname
Standardized
Deviance
Residuals:
targetname
RDT_targetname Studentized
Deviance
Residuals:
targetname
Note: In the table above, targetname is the name of the target variable. For example, if
PURCHASE is the targetname, the predicted value statistic is named P_PURCHA and the
studentized deviance residual is named RDT_PURC. [If the constructed names are longer than the
maximum of eight characters allowed for SAS variable names, they are truncated to eight
characters.]
Statistics Generated in the OUT= SAS-data-set for
Binomial or Multinomial Regression
NAME
LABEL
P_targetname&value
Predicted:
targetname=targetvalue
F_targetname
From: targetname
I_targetname
Into: targetname
E_targetname
Error Function:
R_targetname&value
Residual:
targetname=targetvalue
If the ADDITIONALRESIDUALS option is
specified, the OUT=SAS-data-set includes:
RS_targetname&value Standardized Residual:
targetname=targetvalue
Note: In the table above, targetname&value is a combination of the target name (targetname) and
target value (targetvalue). For example, if PURCHASE is the targetname and "YES" and "NO" are
the two values possible for targetvalue, the predicted value statistics are named P_PURYES and
P_PURNO.
OUTFIT= SAS-data-set
Specifies the output data set with fit statistics. For more information, see .
OUTSTEP
Scores the data for each model selection step.
ROLE=role-value
Specifies the role of the DATA= data set. The ROLE= option primarily affects which fit statistics
are computed and what their names and labels are.
Role-value can be:
TRAIN
This value is the default when the same data set name is used in the DATA= option in both
the PROC and SCORE statements. Specifying TRAIN with any data set other than the
actual training set is an error.
VALID | VALIDATION
This value is the default when the DATA= data set name in the SCORE statement is the
same as the data set in the VALIDATA= in the PROC statement.
TEST
This value is the default when the DATA= data set name in the SCORE statement is the
same as the data set name in the TESTDATA= option of the PROC statement.
SCORE
Predicted values are produced but residuals, error functions, and other fit statistics are not
produced.
Copyright 2000 by SAS Institute Inc., Cary, NC, USA. All rights reserved.