Hands-On Neuroevolution with Python

Data normalization

Often, your input data features have different units of measurement. For example, in the pole-balancing experiment, the cart position was measured in meters, the linear speed was in meters per second, and the angular speed was in radians per second. It is beneficial to normalize input data to simplify the comparison between input data features.

The process of normalization effectively eliminates the units of measurement from the input data samples. After that, all the samples will be in the range between zero and one.

There are different types of normalization in statistics. We already mentioned two methods: data standardization and data range scaling. Additionally, Scikit-learn provides a specialized transformer to perform ...