Neural Network Projects with Python

Data standardization

Data standardization is another important technique in data preprocessing. The goal of data standardization is to transform the numeric variables so that each variable has zero mean and unit variance.

Standardization of variables as a preprocessing step is a requirement for many machine learning algorithms. In neural networks, it is important to standardize the data in order to ensure that the backpropagation algorithm works as intended. Another positive effect of data standardization is that it shrinks the magnitude of the variables, transforming them to a scale that is more proportional.

As we have seen earlier, variables such as Insulin and DiabetesPedigreeeFunction have vastly different scales; the maximum value for ...