Release Notes - IBM SPSS Modeler 16.0 Fix Pack 2

Content

System requirements and installation

For information regarding requirements and installation, see: http://www.ibm.com/software/analytics/spss/products/modeler/requirements.html.

Installing on AIX

LIBPATH=<SP Install Directory>/ext/bin/spss.TMWBServer/bin:$LIBPATH; export LIBPATH

Fixes in this release

The following items have been added in this release:

Scripting API Reference

• Stream class

One new scripting function has been added to the Stream class.

Function: ServerDatabaseConnection - createServerDatabaseConnection (datasourceName, userName, password, catalogName)

Description: Creates a ServerDatabaseConnection. The ODBC data source name must visible to the data mining server.

Parameters:
datasourceName (string): the ODBC data source name
userName (string): the user name
password (string): the password
catalogName (string): the catalog name

Throws a ServerConnectionException if the stream is not connected to a server

Throws a ServerResourceException if the connection attempt fails.

• TaskRunner class

Two new scripting functions have been added to the TaskRunner class.

Function: ModelOutput - importPMMLModelFromFile (filename, autoManage)

Description: Reads PMML from the specified file and returns a model output.

Parameters:
filename (string): the PMML file path
autoManage (boolean): whether the model should be added to the model manager

Throws a SessionException if the model cannot be imported for any reason.




Function: String - publishModelToScoringAdapterTask (node, dbConnection, publishId)

Description: Publishes the model from a model apply node to the specified database
and optional ID. Returns a string representing an example of the SQL needed to call the model.
The node must be connected to another node that provides all the columns necessary
for applying the model. The stream that owns the node must be connected to a server
that can access the database and the database connection must have been
created via the stream that owns the supplied node.

Parameters:
node (Node): the model applier node containing the model
dbConnection (ServerDatabaseConnection): the database connection associated with the node's stream
publishId (string): (DB2 for z/OS databases only) an identifier for the model

Throws a ModelerException if the model cannot be published for any reason.

At time of publication, the following issues were known:

• DB2 password restrictions

When connecting to a DB2 instance, the characters that you cannot use, either on their own or in any combination, are as follows.

`

^

,

;

'

"

(

)

\

&

#

<

>

=

|

$

{

}

In addition, you also cannot use a blank space.



• Windows 32-Bit

Entity Analytics is not supported in local mode on the Win32 platform.



• Paths on UNIX must not include symlinks

On UNIX, the use of symlinks in any paths specified for Entity Analytics will result in an error.



• Entity Analytics installation order

When installing this Fix Pack, you must install the software in the following order:

1 - Modeler 16


2 - Entity Analytics 16
3 - Entity Analytics Unleashed (if you have this additional component)
4 - Modeler 16 FP2
5 - Entity Analytics 16 FP2

• Debug mode displayed when running Entity Analytics nodes

When you start Modeler Solution Publisher from a command window, debug messages can be sent to this window. To avoid this, you can redirect the messages to a file by following these steps:

1. Edit the g2.ini file for the repository and add an entry into the [LOGGING] section.
For example, edit: C:\ProgramData\IBM\SPSS\Modeler\16\EA\repositories\ EAFP2\g2.ini and add an entry called DEBUGCONFIGFILE which points to a file.

[LOGGING]


LOG_OBS=1
DEBUGCONFIGFILE=C:\ProgramData\IBM\SPSS\Modeler\16\ EA\debugInfo.cfg

2. Create the file that you referred to in step 1. (for example debugInfo.cfg) and configure it.

For example, a typical configuration might be:

file://./$NODE_NAME.log *.DBUG;*.CRIT;*.ERR;*.NOTE;*.INFO;*.WARN;*.PERF;


3. Execute the command: "modelerrun.exe -p publishEA.par publishEA.pim"

The debug information that was previously displayed in the command window is written to a file in the C:\ProgramData\IBM\SPSS\Modeler\16\EA\tmp folder.

• Multiple IBM SPSS Modeler installations

There is an error in this section of the IBM SPSS Modeler Text Analytics Server installation instructions. The example of the path is shown as: "/usr/IBM/SPSS/ModelerServer/16".

It should be: "/usr/IBM/SPSS/ModelerServer/16/ext"

• R Nodes and labels

R Nodes do not support variable and value labels; any labels that are included in your data are removed when that data is processed through an R node.

• Differences in binning output

The field operation of binning supports SQL generation if the "Tiles (equal count)" binning method is used and the "Read from Bin Values tab if available" option is selected.

However, due to differences in the way that bin boundaries are calculated (this is caused by the nature of the distribution of data in bin fields), you might see differences in the binning output when comparing normal stream execution results and SQL pushback results. To avoid this, use the Record count tiling method, and either Add to next or Keep in current tiles to obtain the closest match between the two methods of stream execution.

• List datatypes

Lists can be composed only of static objects (for example, a string, number, or field name) and not calls to functions.

To upgrade a Scoring Adapter installation from version 15 to version 16:

1. Complete the SMP/e installation of the version 16 scoring adapter.
2. Create a copy of the resulting SHUMSAMP library and customize it as per the comments in the members. The members are: HUMBIND HUMRACF HUMSCFDB HUMUDFS HUMWLMP HUMWLMA.
3. Update the customized HUMBIND member to free off the existing packages and plans, then bind them and run the job.
4. Update the customized HUMRACF member to delete all racf user and group entries for HUMSPSS and SPSSGRP. In addition, delete the existing started profile for the WLM address space. Then recreate the deleted items and run the job. After the job has completed give the SPSSGRP racf group DBA level access to and authority in your DB2 subsystem, as per your site standards.
5. Run the customized HUMCFDB member.
6. Create the /tmp/cfscoring_16.0.0 directory in USS (permissions 777) and then run the customized HUMUDFS member.
7. Recopy the customized HUMWLMP member to your site's JES2 proclib concatenation to replace the existing version.
8. Quiesce and resume the existing WLM application environment.

Technical Support is available to maintenance customers. Customers may contact Technical Support for assistance in using IBM Corp. products or for installation help for one of the supported hardware environments. To reach Technical Support, see the IBM Corp. web site at http://www.ibm.com/support. We recommend that you check the support site for updates.

© Copyright IBM Corporation 1994, 2015.