INFILE Subcommand (TMS IMPORT command)
The INFILE
subcommand specifies
the location of the XML file containing the transformations.
TRANSFORMATIONS Keyword
The TRANSFORMATIONS
keyword
specifies the location of the XML file containing the transformations.
Specify a filename in quotation marks. The filename must be specified
in full.
The MODE
keyword specifies
the direction of the transformations.
FORWARD
(ROLES= UPDATE|LEAVE). Convert the
"forward" transformations from XML to command syntax. These are the transformations from the original fields into fields
that are used in modeling. ROLES
determines whether to update field roles based upon transformations
in the file. UPDATE
will set
the role of any "obsolete" field to None. LEAVE
will leave the roles of existing fields as they
currently are. FORWARD
is the
default mode. See the topic VARIABLE ROLE for more information. .
BACK(PREDICTED=fieldname
SUFFIX=suffixname). Convert the
backtransformation of target values into command syntax and apply
to the field specified on the PREDICTED keyword. The backtransformed
values are saved to a field with the name on the PREDICTED
keyword and suffix on the SUFFIX
keyword. _backtransformed is the default suffix.
Example
TMS IMPORT
/INFILE TRANSFORMATIONS='file.xml'.
REGRESSION VARIABLES=target_transformed input_transformed
/DEPENDENT=target_transformed
/METHOD=ENTER input_transformed
/SAVE=PRED(predicted) SDBETA(BETA) ICIN.
TMS IMPORT
/INFILE TRANSFORMATIONS='file.xml' MODE=BACK(PREDICTED=predicted).
- The first
TMS IMPORT
reads the file file.xml, converts the transformations from XML to command syntax, and applies them to the active dataset. The fields target_transformed and input_transformed result. - The
REGRESSION
command builds a model for target_transformed using input_transformed, and adds predicted to the active dataset, containing the predicted values. - The second
TMS IMPORT
reads the file file.xml, converts the backtransformations from XML to command syntax, and applies them to the active dataset. The field predicted_backtransformed results, and contains predicted values in the original scale of the target field.