Make project python3 compatible

requirements.txt

@@ -2,3 +2,4 @@
 numpy
 scipy
 scikit-learn
+Cython

skbayes/linear_models/precision_inversion_tester.py

@@ -37,22 +37,22 @@ def inversion_checker(X,alpha,beta):
     beta = 1000
     print('\n Example 1: beta = {0} \n'.format(beta))
     S,v1, v2 = inversion_checker(X, alpha, beta)
-    print "Previous inversion method \n"
+    print("Previous inversion method \n")
     print (v1)
-    print '\n Current inversion method \n'
+    print("\n Current inversion method \n")
     print (v2)
 
     # large beta case
     beta = 1e+16
     print('\n Example 2: beta = {0}  \n'.format(beta))
     S,v1, v2 = inversion_checker(X, alpha, beta)
-    print "Previous inversion method \n"
+    print("Previous inversion method \n")
     print (v1)
-    print '\n Current inversion method \n'
+    print("\n Current inversion method \n")
     print (v2)
 
 
     X = np.random.random([5,5]) + 0.00000001*np.eye(5)
     #print np.linalg.inv(X)
     #print pinvh(X)
-
+

skbayes/linear_models/variational_regression.py

@@ -131,7 +131,7 @@ def fit(self,X,y):
             # --------- Convergence Check ---------
 
             if self.verbose is True:
-                print "Iteration {0} is completed".format(i)
+                print("Iteration {0} is completed".format(i))
 
             # check convergence 
             converged = self._check_convergence(Mw,Mw_old)

skbayes/rvm_ard_models/fast_rvm.py

@@ -596,7 +596,7 @@ def predict_proba(self,X):
         ----------
         X: array-like of size [n_samples_test,n_features]
            Matrix of explanatory variables (test set)
-           
+
         Returns
         -------
         probs: numpy array of size [n_samples_test]
@@ -618,7 +618,7 @@ def predict_proba(self,X):
             prob = pr / np.reshape(np.sum(pr, axis = 1), (pr.shape[0],1))
         return prob
 
-        
+
     def _predict_proba(self,X,y_hat,sigma):
         '''
         Calculates predictive distribution
@@ -627,26 +627,34 @@ def _predict_proba(self,X,y_hat,sigma):
         ks  = 1. / ( 1. + np.pi * var/ 8)**0.5
         pr  = expit(y_hat * ks)
         return pr
-
 
-    def _sparsity_quality(self,X,Xa,y,B,A,Aa,active,Sn):
-        '''
-        Calculates sparsity & quality parameters for each feature
-        '''
-        XB    = X.T*B
-        YB    = y*B
-        XSX   = np.dot(np.dot(Xa,Sn),Xa.T)
-        bxy   = np.dot(XB,y)        
-        Q     = bxy - np.dot( np.dot(XB,XSX), YB)
-        S     = np.sum( XB*X.T,1 ) - np.sum( np.dot( XB,XSX )*XB,1 )
-        qi    = np.copy(Q)
-        si    = np.copy(S) 
-        Qa,Sa      = Q[active], S[active]
-        qi[active] = Aa * Qa / (Aa - Sa )
-        si[active] = Aa * Sa / (Aa - Sa )
-        return [si,qi,S,Q]
-
-
+
+    def _sparsity_quality(self, X, Xa, y, B, A, Aa, active, Sn):
+        '''Calculates sparsity & quality parameters for each feature.'''
+        XB = X.T*B
+        XSX = np.dot(Xa, Sn)
+        XSX = np.dot(XSX, Xa.T)
+
+        S = np.dot(XB, XSX)
+        del XSX
+
+        Q = -np.dot(S, y*B)
+        Q += np.dot(XB, y)
+
+        S *= XB
+        S = -np.sum(S, 1)
+        S += np.sum(XB*X.T, 1)
+        del XB
+
+        qi = np.copy(Q)
+        si = np.copy(S)
+        Qa, Sa = Q[active], S[active]
+        qi[active] = Aa * Qa / (Aa - Sa)
+        si[active] = Aa * Sa / (Aa - Sa)
+
+        return [si, qi, S, Q]
+
+
     def _posterior_dist(self,X,y,A,intercept_prior):
         '''
         Uses Laplace approximation for calculating posterior distribution

skbayes/rvm_ard_models/vrvm.py

@@ -155,11 +155,11 @@ def fit(self,X,y):
 
             # print progress report if required
             if self.verbose is True:
-               print "Iteration {0} is completed, lower bound equals {1}".format(i,self.lower_bound[-1])
+               print("Iteration {0} is completed, lower bound equals {1}".format(i,self.lower_bound[-1]))
 
             if np.sum( abs(Mw - Mw0) > self.tol) == 0 or i == self.n_iter - 1:
                 if self.verbose is True:
-                        print "Mean Field Approximation completed"
+                        print("Mean Field Approximation completed")
                 break
             Mw0 = Mw