Interaction

This transformer takes in Double and Vector type columns and outputs a flattened vector of their feature interactions.

Type

ml-estimator

Class

fire.nodes.etl.NodeInteraction

Fields

Name

Title

Description

inputCols

Input Columns

Column containing label indices

outputCol

Output Column

Output column name

Details

Interaction Node Details

The Interaction Node is used to create a new column with feature interactions between the input columns. It takes in Double and Vector type columns and outputs a flattened vector of their feature interactions. It takes in a DataFrame and transforms it to another DataFrame by adding a new column with the feature interactions.

It takes in the parameters inputCols and outputCol, which are used for input column names and output column name respectively.

Input Parameters

INPUT COLUMNS : Select the required columns for interaction.

OUTPUT COLUMN : The name of the output column after interaction.

Examples

Interaction Node Example

Consider the below Interaction output for the age and income columns.

id age income Interaction

0 20 50000 [20,50000]

1 25 60000 [25,60000]

2 30 70000 [30,70000]

In this example, the input columns are age and income and the output column is Interaction. The interaction node creates a new column with the feature interactions between the input columns. The feature interactions in this case are the concatenation of the values in the columns age and income.