Add a estimate_pairwise_mi method that takes in an array of variables and outputs an MI matrix such that result[i,j] = estimate_mi(vars[i], vars[j]). The user can already implement this method using estimate_mi, but our implementation would offer stronger parallelization.
Should probably return a symmetric matrix while the calculation produces a triangular matrix.
Should have the same parameters for k and conditioning as estimate_mi.
What about lags? Should it be possible to lag one axis?
With no lag, the diagonal could be excluded as auto-MI is usually not useful.
Add a
estimate_pairwise_mimethod that takes in an array of variables and outputs an MI matrix such thatresult[i,j] = estimate_mi(vars[i], vars[j]). The user can already implement this method usingestimate_mi, but our implementation would offer stronger parallelization.kand conditioning asestimate_mi.