This article explains the deployment options for predictions or scoring data with DataRobot. The video shows the different options and explains the benefits of each.
As shown in the video, there are four main methods for deploying a model with DataRobot.
Using the "Make predictions" tab (in the DataRobot GUI) allows you to easily import data for scoring. This option is great for an Excel user doing a quarterly report.
A more automated and real-time scoring method can be accomplished by using the Deploy tab. This tab deploys a model that is available via a REST API. This approach can easily integrate with other IT systems. These deployments can have their system health monitored, data drift evaluated, as well as monitor changes in accuracy.
If you're on Hadoop, DataRobot offers a spark scorer; this is useful for data lakes with large amounts of data.
Finally, you can export scoring code in Java or Python. This can be useful in low latency or offline scoring scenarios.