RECOMMEND Procedure

PREDICT Statement

The PREDICT statement generates recommendations (predictions) for one or more users.

Syntax

Optional Arguments

Syntax

PREDICT / USERLIST=("userID1" <,"userID2" ...>) <option(s)>;

PREDICT / USERDATA=libref.member-name <option(s)>;

Optional Arguments

You can specify the following options in the PREDICT statement after the slash (/):

DATAFILTER="expression"

specifies the optional filter (WHERE clause) for this recommendation. The recommended items are filtered by the expression.

ITEMDATA=table-name

specifies the in-memory table that contains the new transaction data for items. This table can be used when you specify METHOD=ARM to recommend new items based on users' new activities on items.

LABEL='string'

specifies the label that is assigned to the desired method of computing recommendations.

METHOD=AVERAGE | AVE | AVG

METHOD=NEAREST | KNN

METHOD=SLOPEONE | SLOPE1

METHOD=SVD

METHOD=ENSEMBLE

METHOD=ARM

METHOD=CLUSTER

specifies the method to use for computing recommendations. If the requested method is not yet defined for the recommender system, then the SAS LASR Analytic Server adds the method with default parameters and computes the recommendation. For more information, see METHOD Statement.

NRECOMM=n

specifies the upper limit for the number of recommendations that are returned per user.

Alias	NUM=
Default	10

OUT=SAS-data-set

specifies a SAS data set that stores recommendations.

Alias

OUTDATA=

SYSTEM=recommender-system

specifies the name of the recommender system in the SAS LASR Analytic Server that the procedure works with. Specify a two-level name, similar to a LIBNAME.MEMBER construct.

To work with an existing recommender system, the name that you specify identifies the application in the SAS LASR Analytic Server. When you create a new recommender system, the specified value becomes the name of that new system.

Alias	RECOM=
Default	RECOM.SYSTEM

TEMPTABLE

specifies to store the recommendations in a temporary table in the server.

USERDATA=libref.member-name

specifies an in-memory table that contains the user IDs for which you want recommendations. The variable with the user IDs must be named userid.

USERLIST=("userID1" <,"userID2" ...>)

specifies the user IDs for which you want recommendations. This is a convenient format if you want recommendations for only a small number of users.

Alias	USERS=
Interaction	If a value is also specified for USERDATA, then only the USERDATA table is used.