Altair SmartWorks Analytics

 

Features

SmartWorks Analytics presents users with the following features:

General

  • Cloud-based (supports Amazon Web Services and Azure Kubernetes Services)

  • Automated/manual selection of computing engines

  • Easy integration with enterprise environments

  • Editable code behind visual workflow steps

Data Preparation

  • Accurate metrics with auditable change histories and clear data lineage tracking

  • Easy data transformation and preparation

Predictive Analytics and Machine Learning

  • Automatic data preparation, model building, and model comparison using AutoML

  • Prebuilt nodes that allow end-to-end analytic pipelines to be built and executed entirely on:

    • GPUs using NVIDIA’s rapids library for Python

    • Scikit-Learn Python libraries that allow for breadth in functionality

    • SparkML for analytics in big data architectures

Visualization

  • View summary statistics for any given dataframe

  • Generate real-time data previews during data transformation operations

MLOPS

  • Manage and register any Scikit-Learn, Knowledge Studio, Python, TensorFlow, or PySpark code model. Create models in SmartWorks or bring in your own

  • Seamlessly deploy models onto AWS to get a secure endpoint to make predictions, with full control over resources and scaling

  • Feedback Loops – Automatically store model metadata, deployment metadata, and request and response payloads into an external database (PostgreSQL, Oracle, Microsoft SQL Server). Enables full traceability and provenance of models, assessing model performance changes over time and data changes (drift) over time.