Submit your inspiring success story or innovative use case to the 2022 Dataiku Frontrunner Awards! ENTER YOUR SUBMISSION

MATLAB Support

0 Kudos

Working with the scientific community within my company, I've noticed the prevalence of Matlab for implementing data processing and analytics. Matlab excels at the common linear algebra, differential equation, and regression/interpolation modeling needed for scientific data analytics. While R and Python are technically also capable of accomplishing these tasks, they just don't offer the same level of performance or ease of development.

While Matlab's proprietary nature could an obstacle for integration, Matlab Compiler and Matlab Production Server, as well as the Matlab IDE's plugin ecosystem offer a path for integration into DSS. Matlab Online Server also enables a web-IDE that integrates with Production Server and Compiler (all of which are containerized, so standardized deployment in a Dataiku Kubernetes cluster seems relatively straightforward), so a future implementation ofReplace the code editor could potentially support the standard Matlab IDE from directly within the DSS environment. I could also see the potential for integration with Dataiku webapp support, since Matlab has its own data visualization and webapp interface. 

I believe Dataiku could make collaboration and version control much more accessible to scientific users with a Matlab integration, as well as enable these users to become everyday-AI enabled. The ability to simply drop Matlab code directly into the Dataiku environment would make much of the data processing I've been seeing from these users accessible to the broader analytics community as well, eliminating process gaps where data will be sent outside the ecosystem for processing to be returned to Dataiku later.

As a stop-gap, Matlab Production Server can be called from Python to pipe data in, essentially allowing Python to remotely invoke Matlab functions and execute Matlab scripts, but the process of maintaining the Matlab code itself would remain fragmented and obscured from the rest of the analytics community with this approach, where first class citizenship would bring everyone to the same platform. Matlab's userbase seems to be around 4 million, about double that of R's.