During October 2025, Joseph McKenna developed a new variable scaling transformation for the root-project/root repository, enhancing TMVA preprocessing by enabling linear scaling of data to the range [-1, 1] while preserving the sign of input features. He extended the VariableNormalizeTransform class in C++ to support this method, ensuring consistent feature scaling across datasets and facilitating faster convergence for machine learning models. Joseph also updated the associated documentation and usage examples using LaTeX, providing clear guidance for users. This work demonstrated depth in C++ development, data preprocessing, and technical writing, addressing a common challenge in machine learning pipelines.