Trying to fix cv_pls

fluents/lib/validation.py

@@ -61,7 +61,6 @@ def w_pls_cv_val(X, Y, amax, n_blocks=None, algo='simpls'):
     V = w_pls_gen(XXt, Y, n_blocks=n_blocks, center=True)
     for Din, Doi, Yin, Yout in V:
         ym = -sum(Yout, 0)[newaxis]/(1.0*Yin.shape[0])
-        Yin = Yin - ym
         PRESS[:,0] = PRESS[:,0] + ((Yout - ym)**2).sum(0)
         if algo=='simpls':
             dat = w_simpls(Din, Yin, amax)
@@ -74,15 +73,14 @@ def w_pls_cv_val(X, Y, amax, n_blocks=None, algo='simpls'):
         for j in range(l):
             TQ = dot(That, triu(dot(Q[j,:][:,newaxis], ones((1,amax)))) )
             E = Yout[:,j][:,newaxis] - TQ
-            E = E + sum(E, 0)/Din.shape[0]
+            E = E + sum(E, 0)/Din.shape[0] 
             PRESS[j,1:] = PRESS[j,1:] + sum(E**2, 0)
-    Yhat = Y - dot(That,Q.T)
-    rmsep = sqrt(PRESS/Y.shape[0])
-    aopt = find_aopt_from_sep(rmsep)
-    return rmsep, Yhat, aopt
+    #Yhat = Yin - dot(That,Q.T)
+    msep = PRESS/(Y.shape[0])
+    aopt = find_aopt_from_sep(msep)
+    return sqrt(msep)
 
 def pls_val(X, Y, amax=2, n_blocks=10, algo='pls', metric=None):
-    
     k, l = m_shape(Y)
     PRESS = zeros((l, amax+1), dtype='<f8')
     EE = zeros((amax, k, l), dtype='<f8')
@@ -105,9 +103,10 @@ def pls_val(X, Y, amax=2, n_blocks=10, algo='pls', metric=None):
             EE[a,out,:] = E
             PRESS[:,a+1] = PRESS[:,a+1] + sum(E**2,0)
 
-    rmsep = sqrt(PRESS/(k-1.))
-    aopt = find_aopt_from_sep(rmsep)
-    return rmsep, Yhat, aopt
+    #rmsep = sqrt(PRESS/(k-1.))
+    msep = PRESS
+    aopt = find_aopt_from_sep(msep)
+    return msep, Yhat, aopt
 
 def lpls_val(X, Y, Z, a_max=2, nsets=None,alpha=.5):
     """Performs crossvalidation to get generalisation error in lpls"""
@@ -270,7 +269,7 @@ def lpls_jk(X, Y, Z, a_max, nsets=None, alpha=.5):
 def find_aopt_from_sep(sep, method='75perc'):
     """Returns an estimate of optimal number of components from rmsecv.
     """
-
+    sep = sep.copy()
     if method=='vanilla':
         # min rmsep
         rmsecv = sqrt(sep.mean(0))
@@ -282,7 +281,7 @@ def find_aopt_from_sep(sep, method='75perc'):
         med = median(sep)
         prc_75 = []
         for col in sep.T:
-            col.sort()
+            col.sort() #this is inplace -> ruins sep, so we are doing a copy
             prc_75.append(col[int(ind)])
         prc_75 = array(prc_75)
         for i in range(1, sep.shape[1], 1):