Often when analyzing data rescaling is advisable if not a necessity. It’s generally accepted that rescaling a single feature by a strictly monotone function does not cause any loss of information therefore it’s reasonable to conclude that a statistical model based on a dataset \(\mathcal {D}\) is equivalent to a model based on a rescaled data \(\mathcal {D}_1\) . We make the idea of model invariance precise and describe the implications of scaling on various regression and classification models and demonstrate how to interpret a model based on rescaled data.