def two_sample_permu(X0, X1, SigLevel, nSim):
import numpy as np
from permustat import perm_t
XTobs = perm_t.get_T(X0, X1)
XTm = perm_t.get_Ts(X0, X1, nSim)
PvalST = np.zeros((XTm.shape[0], XTm.shape[1]))
for ii in range(XTm.shape[0]):
for tt in range(XTm.shape[1]):
a, b = np.histogram(XTm[ii, tt,:], bins=20)
a = np.cumsum(a)/float(nSim)
if XTobs[ii, tt] < 0:
cutID = np.where(b < XTobs[ii, tt])[0]
if len(cutID)>0 :
PvalST[ii, tt] = a[cutID[-1]]
else:
PvalST[ii, tt] = 0.0
else :
cutID = np.where(b >= XTobs[ii, tt])[0]
if len(cutID)>0 and (cutID[0]<len(a)):
PvalST[ii, tt] = 1.0 - a[cutID[0]]
else:
PvalST[ii, tt] = 0.0
SigST = 1.0 * (PvalST < (SigLevel/2))
nSigS = np.sum(SigST, axis=0)
nSigT = np.sum(SigST, axis=1)
statOut = {'Obs':XTobs,'Pval':PvalST,'Sig':SigST,'nSigS':nSigS,'nSigT':nSigT} #,'XTm':XTm}
return statOut #[XTobs, PvalST, SigST, XTm, nSig, nSigT]
# from scipy.stats import norm
# Xstd = np.std(X, axis=2)
## XT = Xmean / Xstd
# XT = np.sqrt(nX) * Xmean / Xstd
## ut = t.isf(SigLevel/2, nX-1, 0, 1)
# ut = norm.isf(SigLevel/2, 0, 1)
#
# Tvector = XT.reshape((nSpaces*nSamples,1))
# mu = np.mean(Tvector)
# sig = np.std(Tvector)
# vu = mu + ut * sig
# vl = mu - ut * sig
#
# uu = (Tvector > vu )
# ll = (Tvector < vl )
# eee = uu + ll
# eee = eee.reshape((nSpaces, nSamples))
# SigST = 1. * eee
def two_sample_t(X0, X1, SigLevel):
from scipy.stats import t
import numpy as np
from permustat import perm_t
XT = perm_t.get_T(X0, X1)
PvalST = np.zeros(XT.shape)
for ii in range(XT.shape[0]):
for jj in range(XT.shape[1]):
temp = 1-t.cdf(XT[ii,jj], X0.shape[2] + X1.shape[2]-2)
if temp > 0.5 :
PvalST[ii, jj] = 1 - temp
else:
PvalST[ii, jj] = temp
SigST = 1.0 * (PvalST < SigLevel/2)
nSigS = np.sum(SigST, axis=0)
nSigT = np.sum(SigST, axis=1)
statOut = {'T':XT,'Pval':PvalST,'Sig':SigST,'nSigS':nSigS,'nSigT':nSigT}
return statOut #[XT, PvalST, SigST, nSig, nSigT]