Feature Transformation component - ClickML Studio

The Feature Transformation component creates new features derived from selected columns. Unlike Data Transformation (which modifies values in-place), this component adds or replaces columns with newly engineered features.

Configuration

Option	Description	Default
Method	Transformation to apply (see table below)	—
Columns	Columns to use as input for the transformation	—
Degree (Polynomial / Interaction)	Polynomial degree	`2`
N Bins (Binning)	Number of bins	`5`
Bin Strategy (Binning)	`uniform` (equal-width) or `quantile` (equal-frequency)	`uniform`

Methods

Method	What it creates
Polynomial Features	All polynomial combinations up to the given degree (e.g. `x`, `x²`, `x·y`)
Interaction Features	Only interaction terms between selected features (no pure powers)
Binning	Discretizes continuous values into `N` categorical bins
Label Encoding	Encodes string values as integers (in-place)
Frequency Encoding	Replaces each category with its relative frequency in the dataset

Input / Output

	Type
Input	DataFrame
Output	DataFrame