Statistical dependence index

calculates SDI for a matrix of observations

function [argout1 argout2 argout3 argout4]=sdindex2(data,threshold)
%calculation of statistical dependence index
%INPUTS
%data: a two column matrix, each column is a variable, each row an observation
%thresholds: reference thresholds for calculation of probability
%OUTPUTS
%statistical dependence index: SDI = Pab / Pa*Pb
%joint probability: Pab=P(a>threshold & b>threshold)
%A site probability
%B site probability

outputs depend of course on nargout

SDI is presented and defined in

F. Barbaliscia, G. Ravaioli, A. Paraboni. Characteristics of the Spatial Statistical Dependence of Rainfall rates over Large areas. IEEE Trans. Antennas Propagat., Vol. 40, No. 1, pp. 8-12, Feb. 1992

for the context of spatial correlation of rain rate and attenuation for satellite communciations, but is readily extendable to other fields. It is also part of the calculation of mutual information, as log(SDI), so it can help on that.

how to use it:
for a 2-column vector, two sites of n observations, and for a given threshold that you want to evaluate (threshold is used to evaluate probability of exceeding it)

thresholds =[3];
[sdi Pab]=sdindex2(pairdata,thresholds);