pcntoolkit 0.30__tar.gz → 0.30.post2__tar.gz

{pcntoolkit-0.30/pcntoolkit.egg-info → pcntoolkit-0.30.post2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: pcntoolkit
-Version: 0.30
+Version: 0.30.post2
 Summary: Predictive Clinical Neuroscience toolkit
 Home-page: http://github.com/amarquand/PCNtoolkit
 Author: Andre Marquand

{pcntoolkit-0.30 → pcntoolkit-0.30.post2}/pcntoolkit/normative.py

@@ -800,7 +800,11 @@ def predict(covfile, respfile, maskfile=None, **kwargs):
     X = fileio.load(covfile)
     if len(X.shape) == 1:
         X = X[:, np.newaxis]
-
+    if respfile is not None:
+        Y, maskvol = load_response_vars(respfile, maskfile)
+        if len(Y.shape) == 1:
+            Y = Y[:, np.newaxis]
+            
     sample_num = X.shape[0]
     if models is not None:
         feature_num = len(models)
@@ -817,6 +821,11 @@ def predict(covfile, respfile, maskfile=None, **kwargs):
         Xz = scaler_cov[fold].transform(X)
     else:
         Xz = X
+    if respfile is not None:
+        if outscaler in ['standardize', 'minmax', 'robminmax']:
+            Yz = scaler_resp[fold].transform(Y)
+        else:
+            Yz = Y
 
     # estimate the models for all variabels
     for i, m in enumerate(models):
@@ -842,26 +851,16 @@ def predict(covfile, respfile, maskfile=None, **kwargs):
             Yhat[:, i] = yhat.squeeze()
             S2[:, i] = s2.squeeze()
         if respfile is not None:
-            Y, maskvol = load_response_vars(respfile, maskfile)
-            Y = Y[:, i:i+1]
             if alg == 'hbr':
-                if outscaler in ['standardize', 'minmax', 'robminmax']:
-                    Yz = scaler_resp[fold].transform(Y)
-                else:
-                    Yz = Y
-                Z[:,i] = nm.get_mcmc_zscores(Xz, Yz, **kwargs)
-            else:
-                Z[:,i] = (Y - Yhat[:, i]) / np.sqrt(S2[:, i])
-
+                # Z scores for HBR must be computed independently for each model
+                Z[:,i] = nm.get_mcmc_zscores(Xz, Yz[:, i:i+1], **kwargs)
             
-
     if respfile is None:
         save_results(None, Yhat, S2, None, outputsuffix=outputsuffix)
 
         return (Yhat, S2)
-
+    
     else:
-        Y, maskvol = load_response_vars(respfile, maskfile)
         if models is not None and len(Y.shape) > 1:
             Y = Y[:, models]
             if meta_data:
@@ -892,6 +891,10 @@ def predict(covfile, respfile, maskfile=None, **kwargs):
             Y = Yw
         else:
             warp = False
+        
+        if alg != 'hbr':
+            # For HBR the Z scores are already computed 
+            Z = (Y - Yhat) / np.sqrt(S2)
 
         print("Evaluating the model ...")
         if meta_data and not warp:

{pcntoolkit-0.30 → pcntoolkit-0.30.post2/pcntoolkit.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: pcntoolkit
-Version: 0.30
+Version: 0.30.post2
 Summary: Predictive Clinical Neuroscience toolkit
 Home-page: http://github.com/amarquand/PCNtoolkit
 Author: Andre Marquand

{pcntoolkit-0.30 → pcntoolkit-0.30.post2}/setup.py

@@ -14,7 +14,7 @@ requirements = parse_requirements('requirements.txt')
 #       version = '0.29-1'
 
 setup(name='pcntoolkit',
-      version='0.30',
+      version='0.30-2',
       description='Predictive Clinical Neuroscience toolkit',
       url='http://github.com/amarquand/PCNtoolkit',
       author='Andre Marquand',