The JOIN component merges two tables into a single output table by matching records on one or more fields. It supports the same join semantics as SQL: Left Join, Inner Join, Right Join, and Full Join. Use JOIN in the data preprocessing stage of your pipeline to consolidate data from different sources before training or analysis.
Configure the component
Open your pipeline in Machine Learning Designer, then configure the JOIN component in the following order:
Set the join type. Choose how the component handles rows that don't match between the two tables:
Join type Rows included in output Inner Join Only rows that match in both tables Left Join All rows from the left table; unmatched rows from the right table are filled with null Right Join All rows from the right table; unmatched rows from the left table are filled with null Full Join All rows from both tables; unmatched rows on either side are filled with null Add join conditions. Specify the fields to match on as equations. Add or remove conditions as needed.
Select output columns. Choose which columns from the left table and right table to include in the merged output.
Configure MapJoin optimization (optional). If one table is significantly smaller than the other, load it into memory to speed up the join:
Option When to use Not Optimized Both tables are large, or the size difference is small Optimize Left Table The left table is the small table Optimize Right Table The right table is the small table
Parameters
| Parameter | Description |
|---|---|
| Join Type | The join type. Valid values: Left Join, Inner Join, Right Join, Full Join. |
| Join Condition | Join conditions in the format of equations. Add or remove conditions as needed. |
| Select Output Columns from the Left Table | The columns from the left table to include in the output. |
| Select Output Columns from the Right Table | The columns from the right table to include in the output. |
| MapJoin Optimization | Specifies whether to load the small table into memory to accelerate the join. Valid values: Not Optimized, Optimize Left Table, Optimize Right Table. |