Skip to main content

Association analysis

Association analysis (Since version 7.16) measures the strengths of associations between sets of nodes or variables X and Y.

For example, it can be used to measure the strength of links in a network.


Supported for discrete, continuous, hybrid (discrete - continuous) and multi-variate nodes.

Association analysis

Mutual information

Mutual information quantifies the amount of information that can be gained from one variable through the other. Often denoted I(X;Y).


The Entropy for X and Y, denoted H(X) and H(Y) respectively, are reported for convenience, as the normalized forms of the mutual information require them.

Symmetric mutual information

The symmetric mutual information (symmetric uncertainty) is a normalized version of the mutual information. It tells us the strength of the association given the level of uncertainty present in X and Y.

It equals 2 * I(X;Y) / (H(X) + H(Y))


The proficiency (uncertainty coefficient) normalizes the mutual information for either X or Y. For example, the proficiency of X given Y tells us how strong the relationship is given that we know Y.

The proficiency of X given Y, denoted U(X|Y) equals I(X;Y)/H(X), and U(Y|X) equals I(X;Y)/H(Y).