What feature engineering method transforms continuous features by centering their mean around zero and scaling to unit variance?

The feature engineering method that transforms continuous features by centering their mean around zero and scaling to unit variance is known as Standard Scaling. This process involves calculating the z-score for each feature by subtracting the mean of the feature and then dividing by the standard deviation. The result is that the transformed data will have a mean of 0 and a standard deviation of 1, allowing for comparisons across features that may have different units and scales.

Standard Scaling is particularly useful when you are dealing with algorithms sensitive to the scale of data, such as support vector machines and gradient descent-based algorithms, as they perform better when the features are centered and have similar variances.

Min-Max Scaling, on the other hand, adjusts the features to a fixed range, typically between 0 and 1, making it different from Standard Scaling. Robust Scaling focuses on using the median and interquartile range to scale features, which helps with outliers but does not center the data around zero in the same way. Feature Normalization is a more general term that can include various techniques, but it is not specific to the mean and variance transformation that Standard Scaling provides.