def KRR():
np.random.seed(10)
targetValues = np.genfromtxt("F:\PY\data\stock.txt", skip_header=1, dtype=None, delimiter="\t", usecols=(1))
mse = {}
for i in range(20, 50):
trainingPoints = np.arange(i - 1).reshape(-1, 1)
testPoints = np.arange(i).reshape(-1, 1)
# training kernel matrix
knl = mlpy.kernel_gaussian(trainingPoints, trainingPoints, sigma=1)
# testing kernel matrix
knlTest = mlpy.kernel_gaussian(testPoints, trainingPoints, sigma=1)
knlRidge = KernelRidge(lmb=0.01, kernel=None)
err = 0
for j in range(1, 6):
knlRidge.learn(knl, targetValues[-(i + j - 1) : -j])
resultPoints = knlRidge.pred(knlTest)
err += (resultPoints[-1] - targetValues[-j]) ** 2
print i, resultPoints[-1], targetValues[-j]
mse[i] = square(err / 5)
print mse[i]
sortedMse = sorted(mse.iteritems(), key=operator.itemgetter(1), reverse=False)
print sortedMse[0]
def KRR():
np.random.seed(10)
targetValues = np.genfromtxt('F:\PY\data\stock.txt',
skip_header=1,
dtype=None,
delimiter='\t',
usecols=(1))
mse = {}
for i in range(20, 50):
trainingPoints = np.arange(i - 1).reshape(-1, 1)
testPoints = np.arange(i).reshape(-1, 1)
#training kernel matrix
knl = mlpy.kernel_gaussian(trainingPoints, trainingPoints, sigma=1)
#testing kernel matrix
knlTest = mlpy.kernel_gaussian(testPoints, trainingPoints, sigma=1)
knlRidge = KernelRidge(lmb=0.01, kernel=None)
err = 0
for j in range(1, 6):
knlRidge.learn(knl, targetValues[-(i + j - 1):-j])
resultPoints = knlRidge.pred(knlTest)
err += (resultPoints[-1] - targetValues[-j])**2
print i, resultPoints[-1], targetValues[-j]
mse[i] = square(err / 5)
print mse[i]
sortedMse = sorted(mse.iteritems(),
key=operator.itemgetter(1),
reverse=False)
print sortedMse[0]
def KRRTest():
np.random.seed(10)
targetValues = np.genfromtxt('F:\PY\data\stock.txt',
skip_header=1,
dtype=None,
delimiter='\t',
usecols=(1))
trainingPoints = np.arange(46).reshape(-1, 1)
testPoints = np.arange(47).reshape(-1, 1)
#training kernel matrix
knl = mlpy.kernel_gaussian(trainingPoints, trainingPoints, sigma=1)
#testing kernel matrix
knlTest = mlpy.kernel_gaussian(testPoints, trainingPoints, sigma=1)
knlRidge = KernelRidge(lmb=0.01, kernel=None)
knlRidge.learn(knl, targetValues[-46:])
resultPoints = knlRidge.pred(knlTest)
print targetValues.size, resultPoints
fig = plt.figure(1)
plot1 = plt.plot(trainingPoints, targetValues[-46:], 'o')
plot2 = plt.plot(testPoints, resultPoints)
plt.show()
def SmoothKRR():
y = np.genfromtxt('F:\PY\data\Gold.csv',
skip_header=1,
dtype=None,
delimiter=',',
usecols=(1))
targetValues = smooth(y, len(y))
np.random.seed(10)
trainingPoints = np.arange(125).reshape(-1, 1)
testPoints = np.arange(126).reshape(-1, 1)
knl = mlpy.kernel_gaussian(trainingPoints, trainingPoints, sigma=1)
knlTest = mlpy.kernel_gaussian(testPoints, trainingPoints, sigma=1)
knlRidge = mlpy.KernelRidge(lmb=0.01, kernel=None)
knlRidge.learn(knl, targetValues)
resultPoints = knlRidge.pred(knlTest)
print(resultPoints)
plt.step(trainingPoints, targetValues, 'o')
plt.step(testPoints, resultPoints)
plt.show()
def KRR():
np.random.seed(10)
targetValues = np.genfromtxt('F:\PY\data\Gold.csv',
skip_header=1,
dtype=None,
delimiter=',',
usecols=(1))
trainingPoints = np.arange(125).reshape(-1, 1)
testPoints = np.arange(126).reshape(-1, 1)
#training kernel matrix
knl = mlpy.kernel_gaussian(trainingPoints, trainingPoints, sigma=1)
#testing kernel matrix
knlTest = mlpy.kernel_gaussian(testPoints, trainingPoints, sigma=1)
knlRidge = KernelRidge(lmb=0.01, kernel=None)
knlRidge.learn(knl, targetValues)
resultPoints = knlRidge.pred(knlTest)
print(resultPoints)
fig = plt.figure(1)
plot1 = plt.plot(trainingPoints, targetValues, 'o')
plot2 = plt.plot(testPoints, resultPoints)
plt.show()
def SmoothKRR():
y = np.genfromtxt('F:\PY\data\stock.txt',
skip_header=1,
dtype=None,
delimiter='\t',
usecols=(1))
targetValues = smooth(y, len(y))
np.random.seed(10)
trainingPoints = np.arange(28).reshape(-1, 1)
testPoints = np.arange(29).reshape(-1, 1)
knl = mlpy.kernel_gaussian(trainingPoints, trainingPoints, sigma=1)
knlTest = mlpy.kernel_gaussian(testPoints, trainingPoints, sigma=1)
knlRidge = mlpy.KernelRidge(lmb=0.01, kernel=None)
knlRidge.learn(knl, targetValues)
resultPoints = knlRidge.pred(knlTest)
print resultPoints
plt.step(trainingPoints, targetValues, 'o')
plt.step(testPoints, resultPoints)
plt.show()
def klda(X, y):
class_indices = []
class_indices.append(np.where(y == -1)) #at pos 0
class_indices.append(np.where(y == 1)) #at pos 1
#dimension
n = X.shape[0]
d = X.shape[1]
n1 = X[class_indices[0]].shape[0]
n2 = X[class_indices[1]].shape[0]
K = X.dot(X.T)
#Calculate Kernel matrix K1 n1xn1
K1 = mlpy.kernel_gaussian(X, X[class_indices[0]], sigma=15)
K2 = mlpy.kernel_gaussian(X, X[class_indices[1]], sigma=15)
#Calculate means
#Calculate mean matrix M1
M1 = np.zeros((n, 1))
class_no = 0
for i, xi in enumerate(X):
M1[i] = np.sum([[xi.dot(xj.T)] for xj in X[class_indices[class_no]]])
#Calculate mean matrix M2
M2 = np.zeros((n, 1))
class_no = 1
for i, xi in enumerate(X):
M2[i] = np.sum([[xi.dot(xj.T)] for xj in X[class_indices[class_no]]])
#Calculate between class scatter matrix
M = (M2 - M1).dot((M2 - M1).T)
#Calculate within class scatter matrix
N1 = (K1.dot(np.identity(n1) - (np.ones(n1) * n1))).dot(K1.T)
N2 = (K2.dot(np.identity(n2) - (np.ones(n2) * n2))).dot(K2.T)
N = N1 + N2
eig_vals, eig_vecs = np.linalg.eig(np.linalg.inv(N).dot(M))
# Make a list of (eigenvalue, eigenvector) tuples
eig_pairs = [(np.abs(eig_vals[i]), eig_vecs[:, i])
for i in range(len(eig_vals))]
# Sort the (eigenvalue, eigenvector) tuples from high to low
eig_pairs = sorted(eig_pairs, key=lambda k: k[0], reverse=True)
# Construct KxD eigenvector matrix W
W = eig_pairs[0][1]
ldaX = []
for j in range(len(X)):
ldaX.append(
sum([
W[i] * mlpy.kernel_gaussian(X[j], X[i], sigma=15)
for i in range(n)
]))
return ldaX
def f(TrainIn, TrainOut, TestIn): print "init......" x = numpy.array(TrainIn) y = numpy.array(TrainOut) t = numpy.array(TestIn) print "learn......" k = mlpy.kernel_gaussian(x, x, sigma=1) kt = mlpy.kernel_gaussian(t, x, sigma=1) krr = mlpy.KernelRidge(lmb=0.01) krr.learn(k, y) print "out......" re = krr.pred(t) return re
def f(TrainIn, TrainOut, TestIn):
print "init......"
x = numpy.array(TrainIn)
y = numpy.array(TrainOut)
t = numpy.array(TestIn)
print "learn......"
k = mlpy.kernel_gaussian(x, x, sigma=1)
kt = mlpy.kernel_gaussian(t, x, sigma=1)
krr = mlpy.KernelRidge(lmb=0.01)
krr.learn(k, y)
print "out......"
re = krr.pred(t)
return re
def metric(self):
totalTimer = Timer()
with totalTimer:
if self.method_param["kernel"] == "polynomial":
if "degree" in self.method_param:
degree = int(self.method_param["degree"])
else:
degree = 1
kernel = mlpy.kernel_polynomial(self.data[0],
self.data[0],
d=degree)
elif self.method_param["kernel"] == "gaussian":
kernel = mlpy.kernel_gaussian(self.data[0],
self.data[0],
sigma=2)
elif self.method_param["kernel"] == "linear":
kernel = mlpy.kernel_linear(self.data[0], self.data[0])
elif self.method_param["kernel"] == "hyptan":
kernel = mlpy.kernel_sigmoid(self.data[0], self.data[0])
model = mlpy.KPCA()
model.learn(kernel)
out = model.transform(kernel, **self.build_opts)
metric = {}
metric["runtime"] = totalTimer.ElapsedTime()
return metric
def RunKPCAMlpy():
totalTimer = Timer()
# Load input dataset.
Log.Info("Loading dataset", self.verbose)
data = np.genfromtxt(self.dataset, delimiter=',')
try:
with totalTimer:
# Get the new dimensionality, if it is necessary.
if "new_dimensionality" in options:
d = int(options.pop("new_dimensionality"))
if (d > data.shape[1]):
Log.Fatal("New dimensionality (" + str(d) + ") cannot be greater "
+ "than existing dimensionality (" + str(data.shape[1]) + ")!")
return -1
else:
d = data.shape[0]
# Get the kernel type and make sure it is valid.
if not "kernel" in options:
Log.Fatal("Choose kernel type, valid choices are 'polynomial', " +
"'gaussian', 'linear' and 'hyptan'.")
return -1
if options["kernel"] == "polynomial":
if "degree" in options:
degree = int(options.pop("degree"))
else:
degree = 1
kernel = mlpy.kernel_polynomial(data, data, d=degree)
elif options["kernel"] == "gaussian":
kernel = mlpy.kernel_gaussian(data, data, sigma=2)
elif options["kernel"] == "linear":
kernel = mlpy.kernel_linear(data, data)
elif options["kernel"] == "hyptan":
kernel = mlpy.kernel_sigmoid(data, data)
else:
Log.Fatal("Invalid kernel type (" + kernel.group(1) + "); valid " +
"choices are 'polynomial', 'gaussian', 'linear' and 'hyptan'.")
return -1
options.pop("kernel")
if len(options) > 0:
Log.Fatal("Unknown parameters: " + str(options))
raise Exception("unknown parameters")
# Perform Kernel Principal Components Analysis.
model = mlpy.KPCA()
model.learn(kernel)
out = model.transform(kernel, k=d)
except Exception as e:
Log.Fatal("Exception: " + str(e))
return -1
return totalTimer.ElapsedTime()
def RunKPCAMlpy(q):
totalTimer = Timer()
# Load input dataset.
Log.Info("Loading dataset", self.verbose)
data = np.genfromtxt(self.dataset, delimiter=',')
try:
with totalTimer:
# Get the new dimensionality, if it is necessary.
dimension = re.search('-d (\d+)', options)
if not dimension:
d = data.shape[0]
else:
d = int(dimension.group(1))
if (d > data.shape[1]):
Log.Fatal("New dimensionality (" + str(d) + ") cannot be greater "
+ "than existing dimensionality (" + str(data.shape[1]) + ")!")
q.put(-1)
return -1
# Get the kernel type and make sure it is valid.
kernel = re.search("-k ([^\s]+)", options)
if not kernel:
Log.Fatal("Choose kernel type, valid choices are 'polynomial', " +
"'gaussian', 'linear' and 'hyptan'.")
q.put(-1)
return -1
elif kernel.group(1) == "polynomial":
degree = re.search('-D (\d+)', options)
degree = 1 if not degree else int(degree.group(1))
kernel = mlpy.kernel_polynomial(data, data, d=degree)
elif kernel.group(1) == "gaussian":
kernel = mlpy.kernel_gaussian(data, data, sigma=2)
elif kernel.group(1) == "linear":
kernel = mlpy.kernel_linear(data, data)
elif kernel.group(1) == "hyptan":
kernel = mlpy.kernel_sigmoid(data, data)
else:
Log.Fatal("Invalid kernel type (" + kernel.group(1) + "); valid " +
"choices are 'polynomial', 'gaussian', 'linear' and 'hyptan'.")
q.put(-1)
return -1
# Perform Kernel Principal Components Analysis.
model = mlpy.KPCA()
model.learn(kernel)
out = model.transform(kernel, k=d)
except Exception as e:
q.put(-1)
return -1
time = totalTimer.ElapsedTime()
q.put(time)
return time
def KRRProTy():
np.random.seed(10)
targetValues = np.genfromtxt("F:\PY\data\stockPT.txt", skip_header=1, dtype=None, delimiter="\t", usecols=(1))
trainingPoints = np.arange(targetValues.size).reshape(-1, 1)
testPoints = np.arange(targetValues.size + 1).reshape(-1, 1)
# training kernel matrix
knl = mlpy.kernel_gaussian(trainingPoints, trainingPoints, sigma=1)
# testing kernel matrix
knlTest = mlpy.kernel_gaussian(testPoints, trainingPoints, sigma=1)
knlRidge = KernelRidge(lmb=0.01, kernel=None)
knlRidge.learn(knl, targetValues)
resultPoints = knlRidge.pred(knlTest)
print targetValues.size, resultPoints
fig = plt.figure(1)
plot1 = plt.plot(trainingPoints, targetValues, "o")
plot2 = plt.plot(testPoints, resultPoints)
plt.show()
def draw_mlpy_example(data, clabs, testData):
gK = mlpy.kernel_gaussian(data, data, sigma=2) # gaussian kernel matrix
gaussian_pca = mlpy.KPCA(mlpy.KernelGaussian(2.0))
gaussian_pca.learn(data)
gz = gaussian_pca.transform(data, k=2)
fig = plt.figure(1)
ax1 = plt.subplot(121)
plot1 = plt.scatter(data[:, 0], data[:, 1], c=clabs)
plot1_5 = plt.scatter(testData[:, 0], testData[:, 1])
title1 = ax1.set_title('Original data')
trTestData = gaussian_pca.transform(testData, k=2)
ax2 = plt.subplot(122)
plot2 = plt.scatter(gz[:, 0], gz[:, 1], c=clabs)
plot2_5 = plt.scatter(trTestData[:, 0], trTestData[:, 1])
title2 = ax2.set_title('Gaussian kernel')
plt.show()
import numpy as np import matplotlib.pyplot as plt import mlpy np.random.seed(0) x = np.arange(0, 2, 0.05).reshape(-1, 1) # training points y = np.ravel(np.exp(x)) + np.random.normal(1, 0.2, x.shape[0]) # target values xt = np.arange(0, 2, 0.01).reshape(-1, 1) # testing points K = mlpy.kernel_gaussian(x, x, sigma=1) # training kernel matrix Kt = mlpy.kernel_gaussian(xt, x, sigma=1) # testing kernel matrix krr = mlpy.KernelRidge(lmb=0.01) krr.learn(K, y) yt = krr.pred(Kt) fig = plt.figure(1) plot1 = plt.plot(x[:, 0], y, 'o') plot2 = plt.plot(xt[:, 0], yt) plt.show()
""" Confidence level generated from the histogram data """ (con,z_h)=confid_inter.hist_conf_interval(x,y,XHist,YHist) """ Random confidence level """ rcon=confid_inter.conf_interval(len(x),5) """-------Select the histogram confidence level------- """ inter=con """-------Add a third feature of confidence interval to x&y------- """ zc=column_stack((x,y,inter)) z=zc # Adds confidence value to z ######################################################################## """-------Implement kernel PCA with gaussian kernel------- """ xG = mlpy.kernel_gaussian(z, z, sigma=2) # gaussian kernel matrix gaussian_pca = mlpy.KPCA() gaussian_pca.learn(xG) xg = gaussian_pca.transform(xG, k=2) ######################################################################## """-------Implement kernel PCA with polynomial kernel------- """ xP= mlpy.kernel_polynomial(z,z, gamma=1.0, b=1.0, d=2.0) # polynomial kernel matrix polynomial_pca = mlpy.KPCA() polynomial_pca.learn(xP) xp = polynomial_pca.transform(xP, k=3) print 'Implementing Kernel PCA' ########################################################################
def klda(X, y, img_f):
"""
Function to reduce the data and give the reduced transformation matrix
X - The original dimensional data
y - The labels
@returns the dimnsionally reduced data
"""
k = len(np.unique(y))
# Calculate the number of entries for each class
_, Ns = np.unique(y, return_counts=True)
N, m = X.shape
# Obtain all the indices that contain each class separately
class_indices = []
for c in np.unique(y):
class_indices.append(np.where(y == c))
# Calculate the Gram matrix after the Kernel Trick
G = mlpy.kernel_gaussian(X, X, sigma=2.0)
# print G.shape
# Separate the k classes into k different matrices
# Each entry in the c_list is N*nk
c_list = []
te = 0
for i in range(k):
c_temp = G[:, te:te + Ns[i]]
te += Ns[i]
c_list.append(c_temp)
# Initialize the between class scatter matrix and the within class scatter matrix
sb = np.zeros([N, N], np.float32)
sw = np.zeros([N, N], np.float32)
# Calculate the mean of each class
# Each mean vector is N*1
means = []
for i in range(k):
ci = np.sum(c_list[i], 1) / Ns[i]
ci = np.reshape(ci, (N, 1))
means.append(ci)
# Calculate the mean of means
# The mean of means is also a N*1 vector
mean_overall = np.zeros((N, 1), np.float32)
for meani in means:
mean_overall += meani
mean_overall /= k
# Calculate sb
for i in range(k):
sb += Ns[i] * np.matmul((means[i] - mean_overall),
(means[i] - mean_overall).T)
# Calculate sw
for j in range(k):
for i in range(Ns[j]):
sw += np.matmul((c_list[j][:, i] - means[j]),
(c_list[j][:, i] - means[j]).T)
# Calculate the eigen values and sorted eigen vectors of sw_inv_sb
sw_inv_sb = np.matmul(np.linalg.pinv(sw), sb)
eig_vals, eig_vecs = np.linalg.eig(sw_inv_sb)
indices = np.argsort(eig_vals)[::-1]
plot_eigs(eig_vals, indices, img_f)
# Reduce the data
# Choose the dimension to reduce to after analyzing the plot of eigen values
to_red = 4
indices = indices[:to_red]
eig_vecs = eig_vecs[indices]
W = np.reshape(eig_vecs[0], (N, 1))
for i in range(1, to_red):
W = np.concatenate((W, np.reshape(eig_vecs[i], (N, 1))), axis=1)
# print W.shape
return np.matmul(W.T, G)
source = []
target = []
for s in sentences:
words = []
nwords = []
for k in range(len(s) - 1):
words.append(lut[s[k]])
nwords.append(lut[s[k + 1]])
source.append(np.r_[words])
target.extend(np.r_[nwords])
for b in source[:10]:
reservoir.execute(b)
initial_state = reservoir.states[-1]
states = []
for k in range(len(source)):
reservoir.states = np.c_[initial_state].T
states.append(reservoir.execute(source[k]))
X = np.vstack(states)
import mlpy
K = mlpy.kernel_gaussian(X.T, X.T, sigma=1)
readout = mlpy.KernelRidge(lmb=0.01)
readout.learn(K, np.array(target)[:, 0])
kernel_distrib = readout.pred(X)
import numpy as np import os import csv from mlpy import KPCA, kernel_gaussian # Load the data. f = open(os.path.dirname(__file__) + '../data/circle_data.txt') data = np.fromfile(f, dtype=np.float64, sep=' ') data = data.reshape(-1, 2) f.close() # Perform Kernel PCA. kernel = kernel_gaussian(data, data, sigma=2) # gaussian kernel matrix gaussian_pca = KPCA() gaussian_pca.learn(kernel) transformedData = gaussian_pca.transform(kernel, k=2) # Show transformed data. print transformedData
import numpy as np
import matplotlib.pyplot as plt
import mlpy
np.random.seed(0)
mean1, cov1, n1 = [1, 4.5], [[1,1],[1,2]], 20 # 20 samples of class 1
x1 = np.random.multivariate_normal(mean1, cov1, n1)
y1 = np.ones(n1, dtype=np.int)
mean2, cov2, n2 = [2.5, 2.5], [[1,1],[1,2]], 30 # 30 samples of class 2
x2 = np.random.multivariate_normal(mean2, cov2, n2)
y2 = 2 * np.ones(n2, dtype=np.int)
x = np.concatenate((x1, x2), axis=0) # concatenate the samples
y = np.concatenate((y1, y2))
K = mlpy.kernel_gaussian(x, x, sigma=2) # kernel matrix
xmin, xmax = x[:,0].min()-1, x[:,0].max()+1
ymin, ymax = x[:,1].min()-1, x[:,1].max()+1
xx, yy = np.meshgrid(np.arange(xmin, xmax, 0.02), np.arange(ymin, ymax, 0.02))
xt = np.c_[xx.ravel(), yy.ravel()] # test points
Kt = mlpy.kernel_gaussian(xt, x, sigma=2) # test kernel matrix
fig = plt.figure(1)
cmap = plt.set_cmap(plt.cm.Paired)
for i, c in enumerate([1, 10, 100, 1000]):
ka = mlpy.KernelAdatron(C=c)
ax = plt.subplot(2, 2, i+1)
ka.learn(K, y)
ytest = ka.pred(Kt).reshape(xx.shape)
title = ax.set_title('C: %s; margin: %.3f; steps: %s;' % (c, ka.margin(), ka.steps()))
plot1 = plt.pcolormesh(xx, yy, ytest)
plot2 = plt.scatter(x[:,0], x[:,1], c=y)
alpha = np.linalg.solve(N, d) #alpha b = - np.dot(alpha, (n1 * m1 + n2 * m2) / float(n)) # ----- comparare yoonsang's Alpha and package's Alpha ----- import mlpy # mlpy is must be installed with command line function (pip, easy_install) # Using linear kernel Kl = mlpy.kernel_linear(X, X) # compute the kernel matrix linear_kfda = mlpy.KFDA(lmb=0.001) linear_kfda.learn(Kl, y) # compute the tranformation vector zl = linear_kfda.transform(Kl) # embedded x into the kernel fisher space # Using Gaussian kernel sig = 1 Kg = mlpy.kernel_gaussian(X, X, sigma=sig) # compute the kernel matrix gaussian_kfda = mlpy.KFDA(lmb=0.001) gaussian_kfda.learn(Kg, y) # compute the tranformation vector zg = gaussian_kfda.transform(Kg) # embedded x into the kernel fisher space gaussian_kfda._coeff # alpha # Using sigmoid kernel gam=0.1 Ks = mlpy.kernel_sigmoid(X, X, gamma=gam, b=1.0) # compute the kernel matrix sigmoid_kfda = mlpy.KFDA(lmb=0.001) sigmoid_kfda.learn(Ks, y) # compute the tranformation vector zs = sigmoid_kfda.transform(Ks) # embedded x into the kernel fisher space sigmoid_kfda._coeff # Using polynomial kernel gam = 1.0
y = np.convolve(w / w.sum(), s, mode='same')
return y[window_len:-window_len + 1]
y = np.genfromtxt("Gold.csv",
skip_header=1,
dtype=None,
delimiter=',',
usecols=(1))
targetValues = smooth(y, len(y))
np.random.seed(10)
trainingPoints = np.arange(125).reshape(-1, 1)
testPoints = np.arange(126).reshape(-1, 1)
knl = mlpy.kernel_gaussian(trainingPoints, trainingPoints, sigma=1)
knlTest = mlpy.kernel_gaussian(testPoints, trainingPoints, sigma=1)
knlRidge = mlpy.KernelRidge(lmb=0.01, kernel=None)
knlRidge.learn(knl, targetValues)
resultPoints = knlRidge.pred(knlTest)
print(resultPoints)
plt.step(trainingPoints, targetValues, 'o')
plt.step(testPoints, resultPoints)
plt.show()
def klda(X, y):
K = mlpy.kernel_gaussian(X, X, sigma=15)
kfda = mlpy.KFDA(lmb=0.0)
kfda.learn(K, y) # compute the tranformation vector
z = kfda.transform(K) # embedded x into the kernel fisher space
return (z)
y = np.genfromtxt("Gold.csv",
skip_header=1,
dtype=None,
delimiter=',',
usecols=(1))
targetValues = smooth(y, len(y))
np.random.seed(10)
trainingPoints = np.arange(125).reshape(-1, 1)
testPoints = np.arange(126).reshape(-1, 1)
knl = mlpy.kernel_gaussian(trainingPoints, trainingPoints, sigma=1)
knlTest = mlpy.kernel_gaussian(testPoints, trainingPoints, sigma=1)
knlRidge = mlpy.KernelRidge(lmb=0.01, kernel=None)
knlRidge.learn(knl, targetValues)
resultPoints = knlRidge.pred(knlTest)
print(resultPoints)
plt.step(trainingPoints, targetValues, 'o')
plt.step(testPoints, resultPoints)
plt.show()
def RunKPCAMlpy(q):
totalTimer = Timer()
# Load input dataset.
Log.Info("Loading dataset", self.verbose)
data = np.genfromtxt(self.dataset, delimiter=',')
try:
with totalTimer:
# Get the new dimensionality, if it is necessary.
dimension = re.search('-d (\d+)', options)
if not dimension:
d = data.shape[0]
else:
d = int(dimension.group(1))
if (d > data.shape[1]):
Log.Fatal("New dimensionality (" + str(d) +
") cannot be greater " +
"than existing dimensionality (" +
str(data.shape[1]) + ")!")
q.put(-1)
return -1
# Get the kernel type and make sure it is valid.
kernel = re.search("-k ([^\s]+)", options)
if not kernel:
Log.Fatal(
"Choose kernel type, valid choices are 'polynomial', "
+ "'gaussian', 'linear' and 'hyptan'.")
q.put(-1)
return -1
elif kernel.group(1) == "polynomial":
degree = re.search('-D (\d+)', options)
degree = 1 if not degree else int(degree.group(1))
kernel = mlpy.kernel_polynomial(data, data, d=degree)
elif kernel.group(1) == "gaussian":
kernel = mlpy.kernel_gaussian(data, data, sigma=2)
elif kernel.group(1) == "linear":
kernel = mlpy.kernel_linear(data, data)
elif kernel.group(1) == "hyptan":
kernel = mlpy.kernel_sigmoid(data, data)
else:
Log.Fatal(
"Invalid kernel type (" + kernel.group(1) +
"); valid " +
"choices are 'polynomial', 'gaussian', 'linear' and 'hyptan'."
)
q.put(-1)
return -1
# Perform Kernel Principal Components Analysis.
model = mlpy.KPCA()
model.learn(kernel)
out = model.transform(kernel, k=d)
except Exception as e:
q.put(-1)
return -1
time = totalTimer.ElapsedTime()
q.put(time)
return time
def run_stack(SEED):
model = "base"
trainBase = csv_io_np.read_data("PreProcessData/train.csv",
skipFirstLine=True,
split=",")
test = csv_io_np.read_data("PreProcessData/test.csv",
skipFirstLine=True,
split=",")
print "Data Read Complete"
avg = 0
NumFolds = 5
predicted_list = []
bootstrapLists = []
# 100 producted 94%
# 1000 did not finish in about 5+ hours...
# 300 about 5 hours, .9691 on first CF
# learn_rate=0.01, n_estimators=300, subsample=1.0, min_samples_split=30, 0.9386
# GradientBoostingClassifier(loss='deviance', learn_rate=0.1, n_estimators=300, subsample=1.0, min_samples_split=30, min_samples_leaf=1, max_depth=5, init=None, random_state=None, max_features=None)
clfs = [1]
print "Data size: ", len(trainBase), len(test)
dataset_blend_train = np.zeros((len(trainBase), len(clfs)))
dataset_blend_test = np.zeros((len(test), len(clfs)))
trainNew = []
trainTestNew = []
testNew = []
trainNewSelect = []
trainTestNewSelect = []
testNewSelect = []
print "Scaling"
targetPre = [x[0] for x in trainBase]
trainPre = [x[1:] for x in trainBase]
testPre = [x[0:] for x in test]
#print trainPre[0]
#scaler = preprocessing.Scaler().fit(trainPre)
#trainScaled = scaler.transform(trainPre)
#testScaled = scaler.transform(testPre)
trainScaled = trainPre
testScaled = testPre
#print scaler.mean_
#print scaler.std_
print "Begin Training"
lenTrainBase = len(trainBase)
trainBase = []
lenTest = len(test)
test = []
trainPre = []
testPre = []
gc.collect()
CC = [6]
gg = [-6.36, -6.35, -6.34, -6.33, -6.32]
for ExecutionIndex, clf in enumerate(clfs):
print clf
avg = 0
predicted_list = []
dataset_blend_test_set = np.zeros((lenTest, NumFolds))
foldCount = 0
avg = 0
#Stratified for classification...[trainBase[i][0] for i in range(len(trainBase))]
Folds = cross_validation.KFold(lenTrainBase, k=NumFolds, indices=True)
for C in CC:
for g in gg:
for train_index, test_index in Folds:
print "g:", g, "C:", C
#trainBaseTemp = [trainBase[i] for i in train_index]
#target = [x[0] for x in trainBaseTemp]
#train = [x[1:] for x in trainBaseTemp]
#testBaseTemp = [trainBase[i] for i in test_index]
#targetTest = [x[0] for x in testBaseTemp]
#trainTest = [x[1:] for x in testBaseTemp]
#test = [x[0:] for x in test]
target = [targetPre[i] for i in train_index]
train = [trainScaled[i] for i in train_index]
targetTest = [targetPre[i] for i in test_index]
trainTest = [trainScaled[i] for i in test_index]
print
print "Iteration: ", foldCount
print "LEN: ", len(train), len(
train[0]), len(target), len(trainTest), len(
trainTest[0])
K = mlpy.kernel_gaussian(
train, train, sigma=1.0
) # http://elf-project.sourceforge.net/ 26.9048
clf = KernelRidge(
lmb=6.62789e-08) #also might need to set lambda
print datetime.datetime.now()
clf.learn(K, target)
print datetime.datetime.now()
Kt = mlpy.kernel_gaussian(trainTest, train, sigma=1)
prob = krr.pred(Kt)
print datetime.datetime.now()
dataset_blend_train[test_index, ExecutionIndex] = prob
probSum = 0.0
count = 0.0
for i in range(0, len(prob)):
probX = prob[i] #[1]
#print probX, targetTest[i]
if (targetTest[i] == probX):
probSum += 1.0
count = count + 1.0
print "Sum: ", probSum, count
print "Score: ", probSum / count
avg += (probSum / count) / NumFolds
#predicted_probs = clf.predict(testScaled)
######predicted_list.append([x[1] for x in predicted_probs])
#dataset_blend_test_set[:, foldCount] = predicted_probs #[0]
foldCount = foldCount + 1
break
#dataset_blend_test[:,ExecutionIndex] = dataset_blend_test_set.mean(1)
now = datetime.datetime.now()
#csv_io.write_delimited_file_single("../predictions/Stack_" + now.strftime("%Y%m%d%H%M%S") + "_" + str(avg) + "_" + str(clf)[:12] + ".csv", dataset_blend_test_set.mean(1))
#csv_io.write_delimited_file_single("../predictions/Target_Stack_" + now.strftime("%Y%m%d%H%M%S") + "_" + str(avg) + "_" + str(clf)[:12] + ".csv", dataset_blend_train[:,ExecutionIndex] )
csv_io.write_delimited_file("../tune/TuneLog.csv", [
now.strftime("%Y %m %d %H %M %S"), "Score:",
str(avg * NumFolds),
str(clf), "Folds:",
str(NumFolds), "Model", model, "", ""
],
filemode="a",
delimiter=",")
#print "------------------------Average: ", avg
return dataset_blend_train, dataset_blend_test
def run_stack(SEED):
model = "base"
trainBase = csv_io_np.read_data("PreProcessData/train.csv", skipFirstLine = True, split = ",")
test = csv_io_np.read_data("PreProcessData/test.csv", skipFirstLine = True, split = ",")
print "Data Read Complete"
avg = 0
NumFolds = 5
predicted_list = []
bootstrapLists = []
# 100 producted 94%
# 1000 did not finish in about 5+ hours...
# 300 about 5 hours, .9691 on first CF
# learn_rate=0.01, n_estimators=300, subsample=1.0, min_samples_split=30, 0.9386
# GradientBoostingClassifier(loss='deviance', learn_rate=0.1, n_estimators=300, subsample=1.0, min_samples_split=30, min_samples_leaf=1, max_depth=5, init=None, random_state=None, max_features=None)
clfs = [
1
]
print "Data size: ", len(trainBase), len(test)
dataset_blend_train = np.zeros((len(trainBase), len(clfs)))
dataset_blend_test = np.zeros((len(test), len(clfs)))
trainNew = []
trainTestNew = []
testNew = []
trainNewSelect = []
trainTestNewSelect = []
testNewSelect = []
print "Scaling"
targetPre = [x[0] for x in trainBase]
trainPre = [x[1:] for x in trainBase]
testPre = [x[0:] for x in test]
#print trainPre[0]
#scaler = preprocessing.Scaler().fit(trainPre)
#trainScaled = scaler.transform(trainPre)
#testScaled = scaler.transform(testPre)
trainScaled = trainPre
testScaled = testPre
#print scaler.mean_
#print scaler.std_
print "Begin Training"
lenTrainBase = len(trainBase)
trainBase = []
lenTest = len(test)
test = []
trainPre = []
testPre = []
gc.collect()
CC = [6]
gg = [-6.36, -6.35, -6.34, -6.33, -6.32]
for ExecutionIndex, clf in enumerate(clfs):
print clf
avg = 0
predicted_list = []
dataset_blend_test_set = np.zeros((lenTest, NumFolds))
foldCount = 0
avg = 0
#Stratified for classification...[trainBase[i][0] for i in range(len(trainBase))]
Folds = cross_validation.KFold(lenTrainBase, k=NumFolds, indices=True)
for C in CC:
for g in gg:
for train_index, test_index in Folds:
print "g:", g, "C:" , C
#trainBaseTemp = [trainBase[i] for i in train_index]
#target = [x[0] for x in trainBaseTemp]
#train = [x[1:] for x in trainBaseTemp]
#testBaseTemp = [trainBase[i] for i in test_index]
#targetTest = [x[0] for x in testBaseTemp]
#trainTest = [x[1:] for x in testBaseTemp]
#test = [x[0:] for x in test]
target = [targetPre[i] for i in train_index]
train = [trainScaled[i] for i in train_index]
targetTest = [targetPre[i] for i in test_index]
trainTest = [trainScaled[i] for i in test_index]
print
print "Iteration: ", foldCount
print "LEN: ", len(train), len(train[0]), len(target), len(trainTest), len(trainTest[0])
K = mlpy.kernel_gaussian(train, train, sigma=1.0) # http://elf-project.sourceforge.net/ 26.9048
clf = KernelRidge(lmb=6.62789e-08) #also might need to set lambda
print datetime.datetime.now()
clf.learn(K, target)
print datetime.datetime.now()
Kt = mlpy.kernel_gaussian(trainTest, train, sigma=1)
prob = krr.pred(Kt)
print datetime.datetime.now()
dataset_blend_train[test_index, ExecutionIndex] = prob
probSum = 0.0
count = 0.0
for i in range(0, len(prob)):
probX = prob[i]#[1]
#print probX, targetTest[i]
if ( targetTest[i] == probX ) :
probSum += 1.0
count = count + 1.0
print "Sum: ", probSum, count
print "Score: ", probSum/count
avg += (probSum/count)/NumFolds
#predicted_probs = clf.predict(testScaled)
######predicted_list.append([x[1] for x in predicted_probs])
#dataset_blend_test_set[:, foldCount] = predicted_probs #[0]
foldCount = foldCount + 1
break
#dataset_blend_test[:,ExecutionIndex] = dataset_blend_test_set.mean(1)
now = datetime.datetime.now()
#csv_io.write_delimited_file_single("../predictions/Stack_" + now.strftime("%Y%m%d%H%M%S") + "_" + str(avg) + "_" + str(clf)[:12] + ".csv", dataset_blend_test_set.mean(1))
#csv_io.write_delimited_file_single("../predictions/Target_Stack_" + now.strftime("%Y%m%d%H%M%S") + "_" + str(avg) + "_" + str(clf)[:12] + ".csv", dataset_blend_train[:,ExecutionIndex] )
csv_io.write_delimited_file("../tune/TuneLog.csv", [now.strftime("%Y %m %d %H %M %S"), "Score:" , str(avg*NumFolds), str(clf), "Folds:", str(NumFolds), "Model", model, "", ""], filemode="a",delimiter=",")
#print "------------------------Average: ", avg
return dataset_blend_train, dataset_blend_test
source = []
target = []
for s in sentences:
words = []
nwords = []
for k in range(len(s) - 1):
words.append(lut[s[k]])
nwords.append(lut[s[k + 1]])
source.append(np.r_[words])
target.extend(np.r_[nwords])
for b in source[:10]:
reservoir.execute(b)
initial_state = reservoir.states[-1]
states = []
for k in range(len(source)):
reservoir.states = np.c_[initial_state].T
states.append(reservoir.execute(source[k]))
X = np.vstack(states)
import mlpy
K = mlpy.kernel_gaussian(X.T, X.T, sigma=1)
readout = mlpy.KernelRidge(lmb=0.01)
readout.learn(K, np.array(target)[:, 0])
kernel_distrib = readout.pred(X)