Skip to main content

Jump right into Python machine learning with Deephaven

· 2 min read
DALL·E prompt: Blue whale with stacks of shipping containers on it's back, cgsociety artstation trending 4k
JJ Brosnan
Base+ images for AI/ML in Python make getting started quicker than ever

The intersection of AI and real-time data is important. PyTorch, TensorFlow, Sci-Kit-Learn, and NLTK are ubiquitous libraries for data science workflows. To service real-time use cases, they pair well with Deephaven, a query engine that brings Python to "streaming dataframes" -- essentially "tables that update".

Uniting these two powerhouses should be easy. The deephaven.learn library was released in December, providing a gather-compute-scatter paradigm to Deephaven's streaming tables, in support of AI integrations.

Today, the team released an upgrade to the deployment options available to AI-motivated users. With a simple curl script, you can now download a Docker image that combines Python, Deephaven, and the AI library of your choice.

Copy, paste... done.

Last year, we introduced deephaven.learn, the Python module for DS/AI/ML integration with Deephaven tables. We now introduce a suite of base+ Docker images, which prepackage Deephaven with the Python modules you know and love, so that you can start using Deephaven with your favorite modules more easily. Choose from any of the following four popular modules to come pre-packaged with Deephaven:

The project's getting started tutorial and GitHub README provide steps and instructions to start using Deephaven with any of these modules. Choose your deployment, follow the steps, and you'll be up and running in no time.

Are there other Python modules you'd like to see in a base+ image? Let us know on Gitter, GitHub discussions, or file a ticket. We'll be adding to our list of base+ images to accomodate any additional popular modules you'd like to see.

Further reading