Added simple rank correlation option. Only kendalls' approach enabled

mnarayan · Mar 1, 2017 · 26c3dab · 26c3dab
1 parent 48c9360
commit 26c3dab
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 0 deletions.
diff --git a/@GGM/GGM.m b/@GGM/GGM.m
@@ -198,6 +198,7 @@
 
 		[Theta results] = constrainMLE(SigmaHat,InitialTheta,varargin);
 					[W] = adaptiveWeights(Theta,varargin); 
+					Rho = rankCovEstimate(X,varargin); 
 
 	end
 

diff --git a/@GGM/rankCovEstimate.m b/@GGM/rankCovEstimate.m
@@ -0,0 +1,22 @@
+function [Rho] = rankCovEstimate(X,varargin)
+%RANKCOVESTIMATE estimates modified spearman rank correlation as a nonparametric replacement for the sample covariance from standard MLE. 
+% 
+% REFERENCES: 
+% "Regularized Rank Estimation of High Dimensional Nonparanormal Graphical Models", Xue an Zhou (2013)
+% 
+
+
+	% % Use spearman's
+	% Xranks = tiedrank(X,1);
+	% InitRho = corr(Xranks);
+	% Rho  = 2*sin(pi/6.*InitRho);
+
+	% Use Kendall's
+	if(exist('kendalltau') & numel(X)<1e7)
+		Rho = kendalltau(X); 
+	else
+		Rho = corr(X,'type','kendall'); 
+	end
+
+
+end
-Original file line number
+Diff line change
@@ Expand Up / @@ -198,6 +198,7 @@ @@
     		[Theta results] = constrainMLE(SigmaHat,InitialTheta,varargin);
     					[W] = adaptiveWeights(Theta,varargin);
+    					Rho = rankCovEstimate(X,varargin);
     	end
@@ Expand Down @@