Scale of features

Data scaling is a preprocessing technique usually employed before feature selection and classification. Many artificial intelligence-based systems use features that are generated by many different feature extraction algorithms, with  different kinds of sources. These features may have different dynamic ranges. Popular distance measures, for example the Euclidean distance, implicitly assign more weighting to features with large ranges than those with small ranges. Feature scaling is thus required to approximately equalize ranges of the features and make them have approximately the same effect in the computation of similarity.

In addition, in several data mining applications with huge numbers of features with large dynamic ranges, feature scaling may improve the performance of the fitting model. However, the appropriate choice of these techniques is an important issue, since applying scaling on the input could change the structure of data and thereby affect the outcome of...