What's new in IBM SPSS Modeler Subscription?

October 2021

  • An inconvenience with modeling is models getting outdated due to changes to your data over time. This is commonly referred to as model drift or concept drift. To help overcome model drift effectively, SPSS Modeler now provides continuous automated machine learning. A result of IBM research, and inspired by natural selection in biology, continuous machine learning is available for the Auto Classifier node and the Auto Numeric node. For more information, see Continuous machine learning.
  • You can now upload streams to an IBM Cloud Pak for Data server directly from client. For more information, see Saving streams to IBM Cloud Pak for Data.
  • If you want to enable logging for client, open the file log4j2.xml in a text editor and change level="info" to level="debug" in this line:
    <Logger name="com.spss" additivity="false" level="info">
  • A new setting called Perform non negative least squares has been added to the GLE node on the Build Options tab under Parameter Estimation. Non-negative least squares (NNLS) is a type of constrained least squares problem where the coefficients are not allowed to become negative. Not all data sets are suitable for NNLS, which requires a positive or no correlation between predictors and target.
  • A new setting called Lines to scan for column and type has been added to the Excel Source node. The default value for this new setting is 200. If you need to increase this value to scan more rows of your Excel data to determine the column type and storage type, note that performance may be impacted.
  • A new connection method (TM1 Server connection) is available for the IBM Cognos TM1 Source node and the IBM Cognos TM1 Export node to support Planning Analytics on Cloud. The Admin Server has been removed from Planning Analytics on Cloud, so if you have old streams containing TM1 nodes that connect to Planning Analytics on Cloud through TM1 Admin Server, you can now modify them to point to TM1 Server instead.
  • Db2 11.5 is now supported.
  • Db2 Warehouse is now supported.
  • Db2 Big SQL 7.1.0 on Cloudera Data Platform 7.1.5 is now supported.
  • Apache Hive 3.1.3 on Cloudera Data Platform 7.1.5 is now supported.
  • Cloudera Impala 3.4.0 on Cloudera Data Platform 7.1.5 is now supported.
  • Informix 14.10 is now supported.
  • A new version of R is used (4.0.4).
  • For MacOS, the configuration file has been moved out of the application bundle. The new path for the config folder is <installation directory>/Resources/config.

March 2021

  • Oracle Database 19c is now supported.
  • Microsoft SQL Server 2017 is now supported.
  • Vertica 9.3 is now supported.
  • Snowflake is now supported.
  • Plazma database with Treasure Data is now supported, with limitations:
    • You can export data in integer/real format, but not string.
    • Datetime/date/time format isn't supported.
  • Mac Catalina is now supported, version 10.15.3 and above.
  • Previously, the Text Analytics Translate node was deprecated but you could still run existing streams that contained the node. Now, the Translate node is no longer supported.
  • A new version of SDAP is available (8.1).
  • A new version of R is used (3.5.1).
  • A new version of Python is used (3.7.9).

March 2020

New nodes
The following nodes have been added.
  • E-Plot (Beta) node. A new beta-level E-Plot node is available on the Graphs tab of the Nodes palette. It uses a new graphing interface that is intuitive and modern, very customizable, and the data charts are interactive. Use this new node to play around with the new graphing capabilities. For details, see Using an e-plot graph.
  • Gaussian Mixture node. A new Gaussian Mixture node is available on the Python tab and the Modeling tab of the Nodes palette. For details, see Gaussian Mixture node.
  • Kernel Density Estimation (KDE) nodes. A new KDE Modeling node is available on the Python tab and the Modeling tab of the Nodes palette. A new KDE Simulation node is available on the Python tab and the Output tab. For details, see KDE nodes.
  • JSON nodes. New JSON nodes are available for importing and exporting data in JSON format. For details, see JSON Source node and JSON Export node.
New look and feel
A new modern interface theme is available via Tools > User Options > Display. For instructions on switching to the new theme, see Setting display options.
IBM Data Warehouse
Database modeling with IBM Netezza Analytics now supports IBM Data Warehouse. To enable the nodes on the Database Modeling tab in the nodes palette, go to Tools > Options > Helper Applications and enable IBM Data Warehouse integration on the IBM Data Warehouse tab. When you run one of the available Netezza nodes, the built model will now be written to your IBM DB2 Data Warehouse. AIX isn't supported.
IBM SPSS Modeler Text Analytics enhancements
The following enhancements have been made. Most of these enhancements are similar to functionality found in IBM® SPSS® Text Analytics for Surveys .
  • You can now import SPSS Text Analytics for Surveys projects (.tas) in the same way you can import resources from text analysis packages (.tap). When configuring a text mining modeling node, you must specify the resources that will be used during extraction. Instead of choosing a resource template, you can select a .tap or a .tas (new) in order to copy not only its resources but also a category set into the node.
  • Flags are now available in the Data pane. You can flag documents with a "complete" flag or an "important" flag. A new column shows any flags you may be using, and you can click inside the column to change the flag type. This is useful for reviewing the completeness of a category model. See Flagging responses.
  • Extracted concept results have been improved (they're now similar to extracted concept results in SPSS Text Analytics for Surveys )
  • Empty records are now handled the same was as they are in SPSS Text Analytics for Surveys . For example, with an Excel source file, empty records are now kept as part of the text.
  • New Force In and Force Out options are available in the Data pane to force records into or out of a category. This is useful in the case of empty records or records with no extracted concepts, and also when no concept or TLA output enables you to find the appropriate category. See Forcing Responses into Categories.
  • Type Reassignment Rules (TRRs) are now available. TRRs transform a sequence of types, macros, and/or tokens into a new concept with a specific type. They can be used in Opinions templates to catch opinions with a change in polarity. For details, see Type Reassignment Rules.

February 2018

IBM SPSS Modeler Text Analytics is now included with the Subscription offering. IBM SPSS Modeler Text Analytics uses advanced linguistic technologies and Natural Language Processing (NLP) to rapidly process a large variety of unstructured text data, extract and organize the key concepts, and group these concepts into categories. Extracted concepts and categories can be combined with existing structured data, such as demographics, and applied to modeling using the full suite of IBM SPSS Modeler data mining tools to yield better and more focused decisions. For more information about Text Analytics, see the IBM SPSS Modeler Text Analytics Help.

October 2017 (initial IBM SPSS Modeler Subscription offering)

  • Licensing. The IBM SPSS Modeler Subscription licensing process has been replaced by your IBM account, also known as IBMid. An IBMid provides access to all of IBM's applications (to which you are licensed), communities, and support channels. For more information, see Logging on and downloading updates. When you first open IBM SPSS Modeler Subscription, you're prompted to log on with your IBMid. If you don't yet have an IBMid, follow the on-screen instructions.
    Licensing is simplified in IBM SPSS Modeler Subscription. The following licensing options are available:
    Table 1. IBM SPSS Modeler Subscription Licensing Options
    IBM SPSS Modeler Subscription, Base Edition
    IBM SPSS Modeler Subscription, Add-on - SQL Optimization
  • Check for updates. By default, the product automatically informs you whenever new product updates are available. You can choose to either install the updates immediately, or not install the updates. You can also select the Check for Updates option from the Help menu. If you don't want the product to inform you of updates, deselect the Notify on start up when optional updates are available option under Options > User Options > Notifications. For more information, see Logging on and downloading updates.
  • Spark nodes. The new Spark tab on the Nodes palette provides nodes for using Python algorithms. These new nodes are supported on Windows 64 and Mac.
    • Isotonic-AS node. A new Isotonic-AS node is available on the new Spark tab. For details, see Isotonic-AS node.
    • XGBoost-AS node. A new XGBoost-AS node is available on the new Spark tab. For details, see XGBoost-AS node.
    • K-Means-AS node. A new K-Means-AS node is available on the new Spark tab. For details, see K-Means-AS node.
  • Hyper-Parameter Optimization (based on Rbfopt). A new option has been added to the One-Class SVM node (Expert tab), the XGBoost Linear node (Build Options tab), and the XGBoost Tree node (Build Options tab). The new Hyper-Parameter Optimization option automatically discovers the optimal combination of parameters so that the model will achieve the expected or lower error rate on the samples.
  • Random Forest node. A new Random Forest node is available on the Python tab. For details, see Random Forest node.
  • t-SNE node. A new t-Distributed Stochastic Neighbor Embedding (t-SNE) node is available on the Python tab and the Graphs tab. For details, see t-SNE node.
  • Multiple data sources for the CPLEX Optimization node. Optimization experts can now import data from multiple data sources into the CPLEX Optimization node and allocate each data source to a tuple. For details, see CPLEX Optimization node and Settings options for the CPLEX Optimization node.