Usage of discretization and moments

Discretization is one of the most common ways of data transformation. It is a preprocessing step when the applied data mining algorithm requires discrete attributes.

Discretization

Even if the applied data mining algorithm supports continuous attributes, you might consider the following reasons for a data transformation by discretization.

The computational effort of modeling might be considerably reduced.
Usually, the created model is simpler and better readable.
For a classification model, overfitting might be prevented.

Moments

You can use moments for calculating distribution moments and related summary statistics for continuous attributes. Such attributes are, for example, mean, variance, standard deviation, skewness, excess kurtosis, minimum, and maximum.