Table 2 Overview of the normalization methods, their parameters, and the source of implementation
From: The effect of feature normalization methods in radiomics
Method | Parameter | Implementation |
|---|---|---|
z-Score | - | “StandardScaler” from Scikit-learn v1.1.2 |
Robust z-Score (5,95) | Quantile [5,95] | “RobustScaler” from Scikit-learn v1.1.2 |
Robust z-Score (25,75) | Quantile [25,75] | “RobustScaler” from Scikit-learn v1.1.2 |
Min–Max | Scale [-1, 1] | “MinMaxScaler” from Scikit-learn v1.1.2 |
Power transformation | - | “PowerTransformer” from Scikit-learn v1.1.2 |
Quantile transformation | - | “QuantileTransformer” from Scikit-learn v1.1.2 |
Tanh transformation | - | Own implementation |
None | - | - |