A derived view is an asset on the platform that is built using a dataset. The derived view might be a filtered, sorted, or rolled up version of the original dataset. Derived views can be useful for making data more relevant to a specific audience, for reshaping data using grouping, and for controlling access to specific parts of the underlying dataset more granularly.
A derived view can be created on a dataset or on another derived view. A derived view created on another derived view treats the parent derived view as a dataset (i.e. the filter, rollup, and sort conditions are not surfaced in the new derived view). A visualization created using a derived view will surface the filter, sort, and rollup conditions of the derived view it is based on.
On the Enterprise Data Platform, the derived view access model uses a chaining concept to enable finer-grained access control. In this model, if a user with access to a derived view does not have access to the underlying dataset, that user will not know that the underlying dataset exists.
Creating a Derived View
Once you open a dataset in the Exploration Canvas, you can create a new view of that dataset, onto which you can apply filters and shape or join data. To do so, first you will need to click on the Create View button in the upper right corner of the page:
Once you click Create View, you'll be prompted to give the view a name. Once you select the Create View button in the Name your view modal, you'll be able to edit the draft view from within the Exploration Canvas.
Once you've applied your desired filtering/shaping operations, you can click the Publish Dataset button to publish the derived view.