# Measuring the similarity of two distributions

What is the recommended approach to measure the similarity between two distributions of a variable—discrete and continuous? Is Bhattacharyya distance a good measure?

• There are many ways to do this. The best method depends on your application. – user20160 Mar 12 '17 at 13:59
• What about Hellinger, Euclid, Manhatten, Canbera, ...? Try each one out, perform some evaluation (e.g. $k$-fold cross validation), observe the results and pick the metric with the highest result. – Unhandled exception Mar 12 '17 at 15:20