What's new and changed in SPSS Modeler

IBM Cloud Pak for Data Version 5.4.0

A new version of SPSS Modeler was released in June 2026.

This release includes the following changes:

New features

This release of SPSS Modeler includes the following features:

Experiment with TLA rules in the Rule lab

You can now validate and refine your text link analysis (TLA) rules in the Rule lab before applying them to your complete dataset. The Rule lab is an interactive testing environment within the Text Analytics Workbench where you can enter sample text and see how your existing TLA rules match patterns in the sample. When you find patterns that work, you can automatically generate new TLA rules based on the simulation results. With this iterative approach, you can perfect your rules on small samples to save time and improve accuracy before processing large datasets.

For details, see Rule lab.

Updates

The following updates were introduced in this release:

• The SPSS Modeler runtime was upgraded to use the scikit-learn library version 1.5.2. The upgrade addresses security vulnerabilities and ties in with other updates to SPSS Modeler.
• You can use the new runtime environment variable DISABLE_FLIGHT_SQLPUSHBACK to disable the SQL optimization (SQL pushback) when Flight service is enabled. When SQL pushback is disabled, you can still use the Flight service for your databases, but ODBC-compatible data sources default to using ODBC drivers for SQL pushback. For more information, see Runtime environments for SPSS Modeler.
• You no longer need to manually type the data frame setup code in your extension nodes. Now you can use a button to automatically insert the required data frame code, which makes it easier to get started with custom Python or R code.
• You can now sort the rows in the Storage section of the Import node.
• The issues with parsing the Time and Timestamp fields when SQL optimization (SQL pushback) and Flight service are enabled are fixed.

Customer-reported issues fixed in this release

For a list of customer-reported issues that were fixed in this release, see the Fix List for IBM Cloud Pak for Data on the IBM Support website.

Deprecated features

The following features were deprecated in this release:

Older versions of the scikit-learn library are now deprecated

Because SPSS Modeler runtime now uses scikit-learn library version 1.5.2, older versions of the scikit-learn library are deprecated and will be removed in a future release.

This change impacts models that were built with older versions of the scikit-learn library, and it specifically affects models built with the HDBSCAN, KDE, Random Forest (RF), Auto Classifier, and Auto Numeric nodes.

• HDBSCAN and KDE models built with the older versions are not supported. These models fail during scoring, and you'll see error messages in SPSS Modeler. You must rebuild these models by using version 1.5.2.
• Random Forest models built with a version older than scikit-learn library version 1.1.1 are not supported. These models fail during scoring, and you'll see error messages in SPSS Modeler. You must rebuild these models by using version 1.5.2.
• Random Forest models built with a scikit-learn library version between 1.1.1 and 1.5.2 can still be scored, but you'll see warnings in SPSS Modeler. You need to be rebuild these models by using version 1.5.2 if you want to continue using the models.
• Auto Classifier and Auto Numeric nodes are also affected if you selected Random Forest as one of the models in the Expert settings.

Additionally, some specific parameters for nodes are different in scikit-learn library version 1.5.2.

• For KDE nodes, matching and kulsinski were deprecated for the metric parameter.
• For Random Forest nodes, the max_features parameter is now sqrt by default.