def evaluate(self, x, h="silverman"):
density = self.kernel.density
return np.array([density(xx) for xx in x])
if __name__ == "__main__":
PLOT = True
from numpy import random
import matplotlib.pyplot as plt
import bandwidths as bw
# 1 D case
random.seed(142)
x = random.standard_t(4.2, size=50)
h = bw.bw_silverman(x)
#NOTE: try to do it with convolution
support = np.linspace(-10, 10, 512)
kern = kernels.Gaussian(h=h)
kde = KDE(x, kern)
print kde.density(1.015469)
print 0.2034675
Xs = np.arange(-10, 10, 0.1)
if PLOT:
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(Xs, kde(Xs), "-")
ax.set_ylim(-10, 10)
ax.set_ylim(0, 0.4)
#NOTE: I (SS) made some changes to the Fortran
# and the FFT stuff from Munro http://lib.stat.cmu.edu/apstat/97o
# then compile everything and link to denest with f2py
#Make pyf file as usual, then compile shared object
#f2py denest.f -m denest2 -h denest.pyf
#edit pyf
#-c flag makes it available to other programs, fPIC builds a shared library
#/usr/bin/gfortran -Wall -c -fPIC fft.f
#f2py -c denest.pyf ./fft.o denest.f
try:
from denest2 import denest
a = -3.4884382032045504
b = 4.3671504686785605
RANGE = b - a
bw = bandwidths.bw_silverman(xi)
ft, smooth, ifault, weights, smooth1 = denest(xi, a, b, bw,
np.zeros(512),
np.zeros(512), 0,
np.zeros(512),
np.zeros(512))
# We use a different binning algo, so only accurate up to 3 decimal places
np.testing.assert_almost_equal(f2, smooth, 3)
#NOTE: for debugging
# y2 = forrt(weights)
# RJ = np.arange(512/2+1)
# FAC1 = 2*(np.pi*bw/RANGE)**2
# RJFAC = RJ**2*FAC1
# BC = 1 - RJFAC/(6*(bw/((b-a)/M))**2)
# FAC = np.exp(-RJFAC)/BC
#NOTE: I (SS) made some changes to the Fortran
# and the FFT stuff from Munro http://lib.stat.cmu.edu/apstat/97o
# then compile everything and link to denest with f2py
#Make pyf file as usual, then compile shared object
#f2py denest.f -m denest2 -h denest.pyf
#edit pyf
#-c flag makes it available to other programs, fPIC builds a shared library
#/usr/bin/gfortran -Wall -c -fPIC fft.f
#f2py -c denest.pyf ./fft.o denest.f
try:
from denest2 import denest
a = -3.4884382032045504
b = 4.3671504686785605
RANGE = b - a
bw = bandwidths.bw_silverman(xi)
ft,smooth,ifault,weights,smooth1 = denest(xi,a,b,bw,np.zeros(512),np.zeros(512),0,
np.zeros(512), np.zeros(512))
# We use a different binning algo, so only accurate up to 3 decimal places
np.testing.assert_almost_equal(f2, smooth, 3)
#NOTE: for debugging
# y2 = forrt(weights)
# RJ = np.arange(512/2+1)
# FAC1 = 2*(np.pi*bw/RANGE)**2
# RJFAC = RJ**2*FAC1
# BC = 1 - RJFAC/(6*(bw/((b-a)/M))**2)
# FAC = np.exp(-RJFAC)/BC
# SMOOTH = np.r_[FAC,FAC[1:-1]] * y2
# dens = revrt(SMOOTH)
return np.array([self.density(xx) for xx in x])
def evaluate(self, x, h = "silverman"):
density = self.kernel.density
return np.array([density(xx) for xx in x])
if __name__ == "__main__":
PLOT = True
from numpy import random
import matplotlib.pyplot as plt
import bandwidths as bw
# 1 D case
random.seed(142)
x = random.standard_t(4.2, size = 50)
h = bw.bw_silverman(x)
#NOTE: try to do it with convolution
support = np.linspace(-10,10,512)
kern = kernels.Gaussian(h = h)
kde = KDE( x, kern)
print kde.density(1.015469)
print 0.2034675
Xs = np.arange(-10,10,0.1)
if PLOT:
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(Xs, kde(Xs), "-")
ax.set_ylim(-10, 10)