Simplify model creation
You can build and deploy models through our Visual model builder with easy to use wizards or through our integrated Notebooks. This allows data developers and data scientists to focus on the quality of the models and not the complexities of the process.
Easily deploy models
Once created, models can be deployed instantaneously within our framework. RESTful APIs provided in API details allow applications developers to easily incorporate behavioral models into their code.
Easily manage models
The dashboard provides a health check across all models in the enterprise, offering insight into overall model performance and a quick view of those that need to be retrained.
Ensure model accuracy
Data scientists and engineers can schedule continuous re-evaluations on new data to monitor model accuracy over time and be alerted when performance deteriorates.
The z/OS component requires: IBM z/OS, V2.1 or later. The Linux x86 component requires: Red Hat V7.2, or later, OpenJava SDK V1.8.0. The Linux on z component requires: Ubuntu (64-bit) 16.04 or later, OpenSSL 1.0.2g-1ubuntu9.1, openJDK 1.8.0 or later, curl 7.47.0
IBM zEnterprise EC12 or z13 servers running IBM z/OS V2.1 (5650-ZOS), or later. Clients should allocate 4 or more zIIP processors, 1 general purpose processor and 100 GB or more memory to the LPAR where the Program will operate. The Linux component requires x86 hardware and virtual environment OR z13, z13s, zEnterprise EC12, zEnterprise BC12, LinuxOne Emperor, or LinuxOne Rockhopper.
- Standard PMML model support to leverage existing assets
- New MLEAP scoring engine for Spark ML models provide huge performance boost for online scoring
- Feedback data ingestion simplifies model evaluation and feedback loop
- New Administration Dashboard to allow management of spark kernels from web UI