There are multiple ways to change the shape of the data. Usually, it is just projection or filtering, but there are more complex options too.
For row filters, the usual naming convention is used; that is, the node names ending with "Filter" give only a single table as a result, while the "Splitter" nodes generate two tables: one for the matches and one for the non-matching rows.
For single-column conditions, the Row Filter (and Row Splitter) node can be used to select rows based on a column value in a range, regular expression, or missing values. It is also possible to keep only these rows or filter these out. For row IDs, you can only use the regular expressions.
The rows can also be filtered by the (one-based) row index.
The Nominal Value Row Filter node gives a nice user interface when the possible values of textual columns are known at configuration time; so, you do not have to create complex regular expressions to match only those exact values.
There...