Unique

Remove duplicate rows from your data, with the option to output both unique and duplicate rows.

Palette name	Unique (category Compute)
Type name	Unique

Sockets

Socket	Direction	Description
input	Input	Data to deduplicate
output-unique (U)	Output	Rows identified as unique
output-duplicate (D)	Output	Rows identified as duplicates

Two Outputs

Unlike most tools, Unique has two outputs. Connect the U output for unique rows and the D output for duplicates. You can use either or both outputs in your workflow.

Deduplication Modes

Mode	Unique Output	Duplicate Output
First Occurrence	First row of each value group	All subsequent rows
Strictly Unique	Only values appearing exactly once	All rows from repeated groups

First Occurrence (Default)

Groups rows by the selected columns. The first row in each group goes to the unique output; all remaining rows in the group go to the duplicate output.

Strictly Unique

Only rows whose value combination appears exactly once in the dataset go to the unique output. If a value combination appears two or more times, all rows with that combination go to the duplicate output.

Configuration

Option	Default	Description
Columns	(all)	Columns to check for duplicates. Leave empty to check all columns.
Mode	First Occurrence	How to determine which rows are unique

Blocking Operation

Unique is always a blocking operation - it must read all rows before it can determine which are unique and which are duplicates. The entire dataset must fit in memory.

Examples

Remove Duplicate Emails

1. Connect data to the Unique tool
2. Select only the email column
3. Set Mode to First Occurrence
4. Wire the U output to downstream tools

Find All Duplicated Records

1. Connect data to the Unique tool
2. Select the columns that define uniqueness
3. Set Mode to Strictly Unique
4. Wire the D output to see all rows that have duplicates

Deduplicate on Multiple Columns

1. Connect data to the Unique tool
2. Select first_name, last_name, and date_of_birth
3. Set Mode to First Occurrence
4. The first row for each name + DOB combination goes to U, rest to D

Notes

• The output schema for both outputs is identical to the input schema
• When no columns are selected, all columns are used for comparison
• First Occurrence is useful for simple deduplication (keep one, discard rest)
• Strictly Unique is useful for finding records that definitely have no duplicates