Step 3 - Define Dataviews

This step introduces a new structure - Dataviews - structured access to one or more datasets. Users view information in a Highlander database through the dataviews. They form an intermediate layer between the dataset and the user's view of the data.

Dataviews can be transparent and simply pass through a dataset. More often however, dataviews are used to hide dataset dimensions, combine results from several datasets, or make data reductions along a dimension. Built in data reductions include descriptive statistics such as mean and standard deviation, percentiles, and percentile ranges. Custom data reductions can include virtually anything, including calculations using custom algorithms or externally compiled DLL's, if they depend only on the summarized data.

The dataview design opens the database to virtually any data transformation - a powerful OLAP capability. Moreover, it's dynamic. Because dataview transformations are applied at viewing time, there is no need to recompile the database to implement a dataview. Dataviews consume no additional database storage because they reside only in the dictionary as metadata.