Problem

Features whose values exist over a wide scale carry the potential of reducing the predictive potential of features whose values exist over a narrow scale, thereby resulting in the development of a less accurate model.

Solution

All numerical features in a dataset are brought within the same scale so that the magnitude of each feature carries the same predictive potential.

Application

Statistical techniques, such as min-max scaling, mean normalization, and z-score standardization, are applied to convert the features’ values in such a way that the values always exist within a known set of upper and lower bounds.

Mechanisms

Query Engine, Analytics Engine, Processing Engine, Resource Manager, Storage Device

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A training dataset comprises features with varying scales. Feature A’s values are between 0 and 5000, and Feature B’s values are between -15 and +15. Feature C’s values are between 30 and 60 (1). The dataset is exposed to the min-max scaling technique (2). The resulting standardized dataset contains values such that all feature values are now between 0 and 1 (3). The standardized dataset is then used to train a model (4, 5). The resulting model has very high accuracy (6).