pbPassingBI
/
knime/lessons/kn-column-row-filter
Data manipulation beginner 5 min

Column Filter and Row Filter

Reducing the table in both directions.

What you'll be able to do
  • Filter columns by name, type or pattern
  • Filter rows by value or range
  • Split rows into two outputs

Column Filter

Choose columns to include or exclude. Three selection modes:

  • Manual — tick the ones you want
  • Wildcard / Regex — pattern matching, e.g. everything ending _id
  • Type — all numeric columns, all string columns

Wildcard and type selection matter when incoming columns vary — a manual list breaks when a new column appears, a pattern does not.

Filter columns early. Fewer columns means less memory and a faster workflow.

Row Filter

Filters on one column: include or exclude by pattern, by value range, or by row number.

The Row Splitter node is the same thing with two outputs — matching rows from the top port, non-matching from the bottom.

Prefer Row Splitter while developing

Seeing what was excluded is how you catch a filter that removes far more than intended. Once the logic is confirmed you can swap in a Row Filter.

Rule-based filtering

For conditions spanning several columns, Rule-based Row Filter:

$Revenue$ > 1000 AND $Region$ = "East" => TRUE
TRUE => FALSE

Column names go in $ signs. Rules are evaluated in order and the first match wins, so the catch-all belongs last.

Reference filters

Reference Row Filter keeps or removes rows whose key appears in a second table — the equivalent of WHERE id IN (SELECT ...) or an anti-join.

Reference Column Filter does the same for columns, which is how you keep a table's shape consistent with an agreed column list.

Key points
  • Filter columns early — it reduces memory and speeds everything downstream
  • Row Splitter shows what was excluded; Row Filter discards it silently
  • Reference Row Filter is the IN / NOT IN equivalent
Check yourself