The row selector is a component that acts like a visual filter for datasets. It takes one dataset, chops it up into various ranges based on its configuration, and lets the user choose the splices. Then it creates a virtual dataset that only contains the rows that match the selected splices.
The most common way to splice the data is time. You could feed the row selector an input dataset that represents a large time range, and have it break it up by Month, Day, and then Shift, for example. Then you could power a report with the output dataset, and that would let the user dynamically create reports for any time range via an intuitive interface.
To configure the row selector, first you set up the appropriate bindings for its input dataset. Then you use its Customizer to alter the levels that it uses to break up the data. In the customizer, you add various filters that act upon columns in the input dataset, sorting them by various criteria. For example, you could choose a date column, and have it break that up by quarter. Then below that, you could have it use a discrete filter on a product code. This would let the user choose quarterly results for each product. Each level of filter you create in the customizer becomes a level in the selection hierarchy. Note that the output data is completely unchanged other than the fact that rows that don't match the current user selection aren't present.
This component is very handy for driving the Report Viewer, Table, and Classic Chart components, among others.
Additional information on the Row Selector can be found on the Row Selector Component page in the appendix.