How can I normalize mutual information between to real-valued random variables using Python or R? sklearn.metrics.normalized_mutual_info_score seems to work for only nominal data. Or how to interpret the unnormalized scores?


Strictly speaking you cannot just do that as-is.

Mutual information requires knowledge of your data generating probability density function. Nominal data naturally defines such pdf with counts. Continuous data doesn't.

There are two ways to handle that: either assume some pdf, like something parametric (or obtained with kernel smoothing), or convert your data to nominal data (this can be done for example using bucketing).

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.