KNIME today unveiled Integrated Deployment, a new approach aimed at eliminating the gap between the creation of models and their use in production. The announcement was made at the KNIME Spring Summit, taking place virtually now and through April 7. KNIME Integrated Deployment allows not just a model, but all of its associated preparation and post-process steps to be identified and automatically reused in production with no changes or manual work required.
KNIME Analytics is an open source platform for creating data science. It enables the creation of visual workflows via a drag-and-drop-style graphical interface that requires no coding. Users can choose from more than 2000 nodes to build workflows, model each step of analysis, control the flow of data, and ensure work is current. KNIME can blend data from any source and shape data to derive statistics, clean data, and extract and select features. The product leverages AI and machine learning, and can visualize data with classic and advanced charts.
From within the KNIME platform, organizations can replicate this process repeatedly with ease to maintain model performance. Integrated deployment is significant because many business topics that use decision science are affected by this gap. KNIME Integrated Deployment enables the created model, as well as all required steps and settings to be automatically captured and packaged so that the entire production process is available for production use.
Users can create a workflow in KNIME to generate an optimal model. Integrated Deployment allows a data scientist to mark the portions of the workflow that would be necessary for running in a production environment, including data creation and preparation as well as the model itself. There are no limitations in the identification process, and it can be as simple or advanced as needed. With KNIME Server in production, these captured workflows are then referenced and reused.
In a statement to Solutions Review, KNIME CEO and co-founder Michael Berthold told us: “Buyers need to mind the gap when evaluating data science platforms. Until now, there can be huge effort and costs associated with moving data science into production,” said Michael Berthold, CEO and co-founder of KNIME. “KNIME’s Integrated Deployment lets organizations move from creation to deployment automatically, saving time, reducing risk and eliminating those hidden costs. Our users tell us this is a unique approach in the industry.”