Child pages
  • Stratification (7.0)

Contents

Context

Learning | Stratification

Stratification is a tool for modifying the marginal distributions of a subset of variables observed in a data set. It associates internal weights with each observation to reach the user specified marginal distributions. This is particularly useful for variables with very unbalanced distributions.

History

Stratification has been updated in version 5.3 to allow stratifying simultaneously multiple variables.

New Feature: Parameter Estimation

As of version 7.0, the stratification weights can also be taken into account by the Maximum Likelihood Estimation of the parameters. This way, the stratification has an impact not only the structure, but also in the probability distributions.