ncdVSvfmd qualify and leave one out

src/MATLAB/ncdVSvfmd/getStats.m

 
-function tblVelo = getStats(dxx, pData, idxDel, tblMD)
+function tblVelo = getStats(dxx, pData, tblMD)
 if ~exist('idxDel','var')
     idxDel = [];
 end
@@ -9,11 +9,12 @@ for i = 1:length(dxx)
     for j = 1:size(dxx{i},1)
         m = dxx{i}(j,2);
         n = dxx{i}(j,3);
-        if ~isempty(intersect(m,idxDel)) || ~isempty(intersect(n,idxDel))
+        % m and n are indices into pData
+        % check first if either is disqualified
+        if pData.disqualified(m) || pData.disqualified(n)
             continue
         end
        
-       
        if m~=n && pData.dx(m) == pData.dx(n) % same date pairs
             continue
         end

src/MATLAB/ncdVSvfmd/goQualify.m

+load('onh_kernel_corr_pt5_5_15.mat')
+tblQualify = readtable('nyu onh matched.xlsx');
+mx = regexp(tblQualify.file_name,'P(\d+)_(.+?)_(\d+-\d+-\d+)_(\d+-\d+-\d+)_(\w\w)_(.+?)_.*.img','tokens');
+tblQualify.scanID = cellfun(@(x) x{1}{6},mx,'UniformOutput',false);
+
+for i = 1:height(pData)
+
+    idxScan = find(strcmp(tblQualify.scanID,pData.scanID{i})); 
+    qq = lower(vertcat(tblQualify.qualification_status(idxScan)));
+    if length(idxScan) > 1
+        4;
+    end
+    pData.disqualified(i) = any(strcmp(qq,'no'));
+end
+
+

src/MATLAB/ncdVSvfmd/kernelCorr.m

@@ -84,9 +84,7 @@ parfor wx = 1:length(workList)
     end
 end
 
-idxDel = [];
-
-tblVelo = getStats(dxx,pData,idxDel,tblVFMD);
+tblVelo = getStats(dxx,pData,tblVFMD);
 [rho,pCorr] = corr((tblVelo.velocity),(tblVelo.dmd));
 rho2 = rho^2
 toc(kernelCorrStart)

src/MATLAB/ncdVSvfmd/regnetCorr.m

@@ -7,10 +7,22 @@ X1 = [t2.velocity];
 X2 = [t2.velocity,t2.md];
 T = t2.dmd;
 
-mdl1 = fitrnet(X1,T,'Activations','sigmoid','LayerSizes',[1]);
-cv1 = mdl1.crossval; rmse1 = cv1.kfoldLoss^0.5;
-mdl2 = fitrnet(X2,T,'Activations','sigmoid','LayerSizes',[10]);
-cv2 = mdl2.crossval; rmse2 = cv2.kfoldLoss^0.5;
-4;
+% mdl1 = fitrnet(X1,T,'Activations','relu','LayerSizes',[44]);
+% cv1 = mdl1.crossval; rmse1 = cv1.kfoldLoss^0.5;
+% mdl2 = fitrnet(X2,T,'Activations','sigmoid','LayerSizes',[10]);
+% cv2 = mdl2.crossval; rmse2 = cv2.kfoldLoss^0.5;
+% 4;
 % mdl = fitrnet(X,T,"OptimizeHyperparameters","auto", "HyperparameterOptimizationOptions",struct("AcquisitionFunctionName","expected-improvement-plus"))
 
+err = [];
+for i = 1:length(X1)
+    trainX2 = X1;
+    trainX2(i) = [];
+    trainY2 = T;
+    trainY2(i) = [];
+    mdl = fitrnet(trainX2,trainY2,'Activations','sigmoid','LayerSizes',[10]);
+    pred = mdl.predict(X2(i));
+    err(i) = abs(T(i) - pred);
+end
+4;
+