Changed from deprecated scipy.stats.mean() to numpy.mean().

laydi/lib/blmplots.py

@@ -10,7 +10,7 @@ import gtk
 import laydi
 from laydi import plots, main,logger
 import scipy
-from scipy import dot,sum,diag,arange,log,mean,newaxis,sqrt,apply_along_axis,empty
+from scipy import dot,sum,diag,arange,log,newaxis,sqrt,apply_along_axis,empty
 from scipy.stats import corrcoef
 
 def correlation_loadings(data, T, test=True):
@@ -418,7 +418,7 @@ class TRBiplot(plots.ScatterPlot):
         # normalize B
         Bnorm = scipy.apply_along_axis(scipy.linalg.norm, 1, B)
         x = model._dataset['X'].copy()
-        Xc = x._array - mean(x._array,0)[newaxis]
+        Xc = x._array - x._array.mean(0)[newaxis]
         w_rot = B/Bnorm 
         t_rot = dot(Xc, w_rot)
     

laydi/lib/cx_stats.py

@@ -5,7 +5,7 @@ from scipy import zeros,zeros_like,sqrt,dot,trace,sign,round_,argmax,\
      sort,ravel,newaxis,asarray,diag,sum,outer,argsort,arange,ones_like,\
      all,apply_along_axis,eye,atleast_2d,empty
 from scipy.linalg import svd,inv,norm,det,sqrtm
-from scipy.stats import mean,median
+from scipy.stats import median
 
 #import plots_lpls
 
@@ -46,7 +46,7 @@ def hotelling(Pcv, P, p_center='med', cov_center='med',
         P_ctr = median(Pcv, 0)
     elif p_center=='mean':
         # fixme: mean is unstable
-        P_ctr = mean(Pcv, 0)
+        P_ctr = Pcv.mean(0)
     else: #use full
         P_ctr = P
 
@@ -59,7 +59,7 @@ def hotelling(Pcv, P, p_center='med', cov_center='med',
     if cov_center == 'med':
         Cov = median(Cov_i, 0)
     else:
-        Cov = mean(Cov_i, 0)
+        Cov = Cov_i.mean(0)
     
     reg_cov = (1. - alpha)*Cov_i + alpha*Cov
     for i in xrange(n):
@@ -428,7 +428,7 @@ def fdr(tsq, tsqp, loc_method='mean'):
             n_false[j,i] = (tsqp[:,i]>tsq[j]).sum()
     #cPickle.dump(n_false, open("/tmp/nfalse.dat_"+str(n), "w"))
     if loc_method=='mean':
-        fp = mean(n_false,1)
+        fp = n_false.mean(1)
     elif loc_method == 'median':
         fp = median(n_false.T)
     else:

laydi/lib/cx_utils.py

@@ -3,7 +3,7 @@ from scipy import apply_along_axis,newaxis,zeros,\
      trace,zeros_like,sign,sort,real,argsort,rand,array,\
      matrix,nan
 from scipy.linalg import norm,svd,inv,eig
-from scipy.stats import median,mean
+from scipy.stats import median
 
 def normalise(a, axis=0, return_scales=False):
     s = apply_along_axis(norm, axis, a)
@@ -99,11 +99,11 @@ def mat_center(X,axis=0,ret_mn=False):
         print "The X data needs to be two-dimensional"
         
     if axis==0:
-        mnX = mean(X,axis)[newaxis]
+        mnX = X.mean(axis)[newaxis]
         Xs = X - mnX
     
     elif axis==1:
-        mnX = mean(X,axis)[newaxis]
+        mnX = X.mean(axis)[newaxis]
         Xs = (X.T - mnX).T
     if ret_mn:
         return Xs,mnX

laydi/lib/select_generators.py

@@ -1,6 +1,6 @@
 """Matrix cross validation selection generators
 """
-from scipy import take,arange,ceil,repeat,newaxis,mean,asarray,dot,ones,\
+from scipy import take,arange,ceil,repeat,newaxis,asarray,dot,ones,\
      random,array_split,floor,vstack,asarray,minimum
 from cx_utils import randperm
 
@@ -188,7 +188,7 @@ def diag_pert(a, n_sets=10, center=True, index_out=False):
     nm=n*m
     start_inds = array_split(randperm(m),n_sets) # we use random start diags
     if center:
-         a = a - mean(a, 0)[newaxis]
+         a = a - a.mean(0)[newaxis]
     for v in range(n_sets):
         a_out = a.copy()
         out = []

laydi/lib/validation.py

@@ -1,6 +1,6 @@
 """This module implements some common validation schemes from pca and pls.
 """
-from scipy import ones,mean,sqrt,dot,newaxis,zeros,sum,empty,\
+from scipy import ones,sqrt,dot,newaxis,zeros,sum,empty,\
      apply_along_axis,eye,kron,array,sort,zeros_like,argmax,atleast_2d
 from scipy.stats import median
 from scipy.linalg import triu,inv,svd,norm