SPSS Modeler

 View Only

What's New in SPSS Modeler - Cloud Pak for Data 4.5 Release

By Jacob Stellon posted Wed June 29, 2022 06:19 PM

When Cloud Pak for Data 4.5 is released at the end of June 2022, SPSS Modeler users will notice an impressive number of changes that help users boost efficiency and give advanced control to experts. Data scientists can use SPSS Modeler in Cloud Pak for Data to prepare data and build machine learning models without code in an iterative, rapid fashion. Whether it's to find the right algorithm or to experiment with data, SPSS Modeler users can create reproducible research that's easily understood by any member of their team.

The Cloud Pak for Data 4.5 release includes the following feature enhancements:
  1. Node generation from tables
  2. Text Analytics vNext and Resource Editor
  3. Improved scripting experience
  4. SQL pushback preview
  5. Output comparison view
  6. XGBoost Tree Model Viewer

1. Node Generation from Tables
Users can quickly dive into specific segments of their data by generating Select and Derive nodes from a table. Generate a Select node to hone in on one portion of the data or a Derive node to create a new field without the need for code. 

2. Text Analytics vNext and Resource Editor
The Text Mining node allows users to create actionable business insights from their unstructured text data. In CPD 4.5 the SPSS Modeler team enhanced the interface and added the Resource Editor to give advanced users precise control over their linguistic models while maintaining the ease of use for which the product is known. Users can easily create distribution charts to visualize themes in text data or use Text Analytics to improve a predictive model. Now, it's easier than ever to modify a concept or category model with domain knowledge by adding synonyms or creating advanced text link analysis rules. The Resource Editor gives experts the flexibility they need to fine tune their model. The following videos highlight the enhancements to the Text Mining node through the use cases of understanding survey data and improving a predictive model.

3. Improved Scripting Experience​
Scripting in SPSS Modeler is a powerful tool for automating processes in the user interface. Scripts can perform the same types of actions that users perform with a mouse or a keyboard, and scripts are useful for automating tasks that would be highly repetitive or time consuming to perform manually. The redesigned scripting field includes syntax highlighting, line numbering, block matching, and suggested auto-complete to give users a notebook like experience directly in the product. Read the scripting documentation or watch a video demoing three scripting use cases to get started with this advanced capability.

4. SQL Pushback Preview
Performance is a crucial consideration for SPSS Modeler users. Because of this, SPSS Modeler in CPD automatically pushes many data preparation and mining operations directly to supported databases. To further optimize, users can strategically build their flows by taking actions like moving non-pushback operations further downstream. New in the Cloud Pak for Data 4.5 release, users can preview which nodes will have SQL pushed back to a connected database allowing them to easily optimize their flow for performance. Read more about SQL pushback here.

5. Compare Outputs
Whether it's to dive deep into a segment of data, discover a trend, or evaluate competing models, efficiency is key. Users can now quickly compare outputs like charts, tables, and model metrics to speed up the time to insight while building a flow.

6. XGBoost Tree Model Viewer
Explainable machine learning models are table stakes for building trust in AI. In Cloud Pak for Data 4.5 the SPSS Modeler team has added the XGBoost Tree model viewer which displays evaluation metrics, model information, feature importance, and a confusion matrix so data scientists can easily understand their model after building an XGBoost Tree Model directly or in an Auto Modeling node. 

The Cloud Pak for Data 4.5 release marks an important milestone in the continued development of SPSS Modeler. As the team looks to what's next, we hope to continue building on the solid history of the SPSS Modeler product while providing innovating new features.