Data Science

Provides a brief overview of data science in HPE Ezmeral Unified Analytics Software.

Data scientists can use programming languages such as Python, R, Java, and SQL to build, train, and deploy machine learning models in HPE Ezmeral Unified Analytics Software using open-source tools that optimize the performance of predictive machine learning models.

Data scientists can use the tools provided in HPE Ezmeral Unified Analytics Software to:
  • Perform exploratory data analysis in Notebooks.
  • Build features or labels from the data.
  • Create and train models in Notebooks or Pipelines and training frameworks like TensorFlow, Ray, or PyTorch.
  • Create and run pipelines based on variable conditions for repetitive tasks.
  • Run jobs across the distributed clusters or cloud burst (launch) the jobs into a separate cloud environment using APIs from Kubeflow.
  • Select your model and hyperparameters for your model to run AutoML jobs by using Katib and MLflow.
  • Compile the models into a container and enter the container into the registry to make it available for model serving as a part of KServe.
  • Query pipelines for data drift, bias, and robustness.
  • Evaluate models and replace the previous models for optimization or retrain and deploy the models for better performance.