Dimension hierarchies

Cloudera Data Visualization provides full support for dimensional hierarchy modelling at dataset level, enabling smooth natural transition between granularity levels of data.

Hierarchies are a natural form of data organization. For example, you can break up any given time interval into intervals of shorter duration based on years, months, weeks, days, and so on. Similarly, the territory of United States consists of well-defined states, then counties, then municipalities and incorporated and unincorporated areas.

Hierarchies may only be defined through columns that are classified as dimensions. If necessary, change the column attribute from Measure to Dimension. Alternatively, clone a column, change it to a dimension, and only then use it in defining a dimensional hierarchy.

Just like with column names, the name of the dimensional hierarchy must be unique to the dataset.

Columns that define a dimension hierarchy are retained as a reference, not a copy. Therefore, changes to the basic definition of the column (such as name, type, calculation) propagate into the dimensional hierarchy.

To build out dimensional hierarchies, use 'drag and drop' action to add dimension columns, and to prioritize them. Cloudera recommends that you use the delete function to permanently remove items from a hierarchy. You can also move elements from one hierarchy to another; note that this action removes the dimension from the source hierarchy and adds it to the target hierarchy.