Y-scaling refers to the process of adjusting the scale of the target variable, or output (Y), in a machine learning model. This process is often used to normalize or standardize the Y-values to a common scale, which can improve the performance and convergence of models, especially in regression tasks. The meaning of y-scaling is tied to its role in ensuring that the model's predictions are on the same scale as the actual outputs, which can be crucial for accurate and interpretable results.
Y-true, also known as actual output, refers to the true or observed values in a dataset that a machine learning model aims to predict. These values are the ground truth against which the model's predictions, known as y-pred or predicted output, are compared. The meaning of y-true is central to the evaluation of a model's accuracy, as it represents the correct outcomes that the model should strive to replicate.