DataOps and MLOps as machine learning problem solvers

BARC study
DataOps and MLOps as machine learning problem solvers

providers on the subject

The Business Application Research Center (BARC) conducted an international study around the perception of DataOps and MLOps. It shows the problems addressed and the effects of these concepts in companies.

The BARC study shows advantages of DataOps and MLOps.

(Image: BARC )

The study entitled “Driving Innovation with AI. Getting Ahead with DataOps and MLOps” was created in cooperation with ONE LOGIC, Domino Data Lab and DataRobot. 248 companies of different sizes and industries were surveyed for them.

Results at a glance

As the BARC study shows, DataOps and MLOps are widely recognized as approaches to address common machine learning problems and challenges. However, only half of all companies that are already using machine learning use it. Almost all (97 percent) of companies already using DataOps and MLOps were able to achieve significant improvements as a result. This includes, for example, the easier handling of challenges and complexity as well as the faster provision of models.

Open source dominates the implementation of machine learning. According to the BARC, commercial tools and platforms could reduce complexity and simplify and accelerate the delivery of results. Numerous positive side effects were also recorded in the implementation of DataOps and MLOps, for example in the recruitment of specialists or the use of new tools, platforms and infrastructures. However, central challenges such as data silos or promoting the general acceptance of machine learning remain.

Make processes more agile and efficient

“DataOps is aimed at realizing a manageable, maintainable and automated flow of quality-assured data to data products,” explains Alexander Rode, Data & Analytics Analyst at BARC and co-author of the study. MLOps address the additional special requirements regarding the development, deployment and maintenance of machine learning models.

“The aim of both concepts in this context is to create transparency about all dependencies between the systems involved along an end-to-end data pipeline and to promote collaboration between experts by simplifying the process of developing, updating and maintaining ML models are made more agile and efficient,” Rode continues.


#DataOps #MLOps #machine #learning #problem #solvers

Leave a Comment

Your email address will not be published.