User manual SAS ANALYTICS ACCELERATOR 1.3 FOR TERADATA

[. . . ] SAS Analytics Accelerator 1. 3 for Teradata Guide ® SAS® Documentation The correct bibliographic citation for this manual is as follows: SAS Institute Inc. SAS® Analytics Accelerator 1. 3 for Teradata: Guide. SAS® Analytics Accelerator 1. 3 for Teradata: Guide Copyright © 2011, SAS Institute Inc. , Cary, NC, USA All rights reserved. Produced in the United States of America. [. . . ] DATABASE= as a data set option does not work like DATABASE= as a LIBNAME option. For out-of-database I/O. DBCOMMIT= DBCONDITION= DBCREATE_TABLE_OPTS= DBFORCE= N/A No3 N/A N/A N/A No3 Yes1 N/A In-database computation works like DBFORCE=YES, regardless of what value is specified for DBFORCE=. SAS/ACCESS software defaults to DBFORCE=NO. However, DBFORCE= would have an effect on in-database computation only if DBTYPE= specified a Teradata data type that resulted in truncation, whereas specifying DBTYPE= prevents in-database computation. For out-of-database I/O. DBINDEX= DBKEY= N/A N/A N/A N/A SAS/ACCESS Data Set Options for Teradata !33 Data Set Option DBLABEL= DATA= N/A OUT= N/A Comments DBLABEL= applies only when the DATA= data set is a SAS data set with variable labels and the OUT= data set is a DBMS table. In-database computation cannot be performed when the DATA= data set is not a DBMS table. DBMASTER= DBNULL= DBSASLABEL= N/A N/A3 Yes1 N/A No3 N/A DBSASLABEL=NONE is recommended for correct procedure output, but it is normally easier to specify this in the LIBNAME statement. For out-of-database I/O. DBSLICE= DBSLICEPARM= DBTYPE= N/A N/A N/A3 N/A N/A Sometimes3 DBTYPE= prevents in-database computation because some Teradata types such as DATE and TIMESTAMP have not been implemented. N/A N/A N/A N/A N/A N/A N/A N/A N/A For out-of-database I/O. For out-of-database I/O. ERRLIMIT= MBUFSIZE= ML_CHECKPOINT= ML_ERROR1= ML_ERROR2= ML_LOG= ML_RESTART= ML_WORK= MULTILOAD= N/A N/A N/A N/A N/A N/A N/A N/A N/A 34 !Chapter 1: SAS Analytics Accelerator 1. 3 for Teradata: Guide Data Set Option MULTISTMT= NULLCHAR= NULLCHARVAL= PRESERVE_COL_NAMES=NO READ_ISOLATION_LEVEL= READ_LOCK_TYPE= READ_MODE_WAIT= SCHEMA= SET= DATA= N/A N/A3 N/A3 No3 No4 No4 No4 Yes1 N/A OUT= N/A No3 No3 No3 No4 No4 No4 Yes1 Yes2 Comments For out-of-database I/O. The default value YES works. The option is ignored. If the DATA= table contains duplicate rows, the procedure attempts to insert duplicate rows into the OUT= table, and the results for in-database and out-of-database processing differ. With in-database processing, duplicate rows are discarded and do not cause errors. With out-of-database processing, duplicate rows cause errors and a ROLLBACK is issued. With MODE=TERADATA, the ROLLBACK produces an empty or incomplete table. With MODE=ANSI, the ROLLBACK produces an empty table. The option is ignored. SLEEP= TENACITY= UPDATE_ISOLATION_LEVEL= UPDATE_LOCK_TYPE= UPDATE_MODE_WAIT= N/A N/A No4 No4 No4 N/A N/A No4 No4 No4 Base SAS Data Set Options !35 Base SAS Data Set Options Table 1. 6 describes whether the Base SAS data set options work with the DATA= and OUT= SAS/STAT and SAS/ETS procedure options for in-database computing. The cell entries in the DATA= and OUT= columns have the following values: Yes--indicates that the option works for in-database computing No--indicates that the option does not work for in-database computing N/A (not applicable)--indicates that the option does not apply to in-database computing Some cell entries in the DATA= and OUT= columns have superscripts that are intended to convey additional information. The meanings of the superscripts are as follows: A superscript of 1 (Yes1 or No1 ) indicates an option that correctly affects in-database processing as expected. A superscript of 2 (Yes2 , No2 , or N/A2 ) indicates an option that causes SAS/ACCESS syntax errors or warnings. When SAS/ACCESS software issues a warning message, the option is ignored by the procedure. [. . . ] SAS output provides clues about how the factors are used in the model. title3 'Linear Regression on raw data in teradata'; ods select SelectionSummary; proc reg data=tera. tdview_reg; model target = age amount checking coapp depends duration employed existcr history housing installp job marital other property foreign resident savings telephon / selection = stepwise; run; quit; Example !69 8 Suppose you want to run several model statements, selecting different combinations of variables and options. You can use the CORR matrix that you saved from the output of the PROC PRINCOMP step. PROC CORR and PROC REG can also produce this matrix as output. [. . . ]