y = dataset['y']
Xtrain, Xtest, ytrain, ytest = train_test_split(X, y, random_state=0)
# standard logistic regression
lr = logRegL2(lammy=1)
lr.fit(Xtrain, ytrain)
print("Training error %.3f" % np.mean(lr.predict(Xtrain) != ytrain))
print("Validation error %.3f" % np.mean(lr.predict(Xtest) != ytest))
utils.plotClassifier(lr, Xtrain, ytrain)
utils.savefig("logReg.png")
# kernel logistic regression with a linear kernel
lr_kernel = kernelLogRegL2(kernel_fun=logReg.kernel_linear, lammy=1)
lr_kernel.fit(Xtrain, ytrain)
print("Training error %.3f" %
np.mean(lr_kernel.predict(Xtrain) != ytrain))
print("Validation error %.3f" %
np.mean(lr_kernel.predict(Xtest) != ytest))
utils.plotClassifier(lr_kernel, Xtrain, ytrain)
utils.savefig("logRegLinearKernel.png")
elif question == "1.1":
dataset = load_dataset('nonLinearData.pkl')
X = dataset['X']
y = dataset['y']
y = dataset['y']
Xtrain, Xtest, ytrain, ytest = train_test_split(X, y, random_state=0)
# standard logistic regression
lr = logRegL2(lammy=1)
lr.fit(Xtrain, ytrain)
print("Training error %.3f" % np.mean(lr.predict(Xtrain) != ytrain))
print("Validation error %.3f" % np.mean(lr.predict(Xtest) != ytest))
utils.plotClassifier(lr, Xtrain, ytrain)
utils.savefig("logReg.png")
# kernel logistic regression with a linear kernel
lr_kernel = kernelLogRegL2(kernel_fun=logReg.kernel_linear, lammy=1)
lr_kernel.fit(Xtrain, ytrain)
print("Training error %.3f" %
np.mean(lr_kernel.predict(Xtrain) != ytrain))
print("Validation error %.3f" %
np.mean(lr_kernel.predict(Xtest) != ytest))
utils.plotClassifier(lr_kernel, Xtrain, ytrain)
utils.savefig("logRegLinearKernel.png")
elif question == "1.1":
dataset = load_dataset('nonLinearData.pkl')
X = dataset['X']
y = dataset['y']
y = dataset['y']
Xtrain, Xtest, ytrain, ytest = train_test_split(X,y,random_state=0)
# standard logistic regression
lr = logRegL2(lammy=1)
lr.fit(Xtrain, ytrain)
print("Training error %.3f" % np.mean(lr.predict(Xtrain) != ytrain))
print("Validation error %.3f" % np.mean(lr.predict(Xtest) != ytest))
utils.plotClassifier(lr, Xtrain, ytrain)
utils.savefig("logReg.png")
# kernel logistic regression with a linear kernel
lr_kernel = kernelLogRegL2(kernel_fun=logReg.kernel_linear, lammy=0.01)
lr_kernel.fit(Xtrain, ytrain)
print("Training error %.3f" % np.mean(lr_kernel.predict(Xtrain) != ytrain))
print("Validation error %.3f" % np.mean(lr_kernel.predict(Xtest) != ytest))
utils.plotClassifier(lr_kernel, Xtrain, ytrain)
utils.savefig("logRegLinearKernel.png")
elif question == "1.1":
dataset = load_dataset('nonLinearData.pkl')
X = dataset['X']
y = dataset['y']
Xtrain, Xtest, ytrain, ytest = train_test_split(X,y,random_state=0)
y = dataset['y']
Xtrain, Xtest, ytrain, ytest = train_test_split(X, y, random_state=0)
# standard logistic regression
lr = logRegL2(lammy=1)
lr.fit(Xtrain, ytrain)
print("Training error %.3f" % np.mean(lr.predict(Xtrain) != ytrain))
print("Validation error %.3f" % np.mean(lr.predict(Xtest) != ytest))
utils.plotClassifier(lr, Xtrain, ytrain)
utils.savefig("logReg.png")
# kernel logistic regression with a linear kernel
lr_kernel = kernelLogRegL2(kernel_fun=logReg.kernel_linear, lammy=1)
lr_kernel.fit(Xtrain, ytrain)
print("Training error %.3f" %
np.mean(lr_kernel.predict(Xtrain) != ytrain))
print("Validation error %.3f" %
np.mean(lr_kernel.predict(Xtest) != ytest))
utils.plotClassifier(lr_kernel, Xtrain, ytrain)
utils.savefig("logRegLinearKernel.png")
elif question == "1.1":
dataset = load_dataset('nonLinearData.pkl')
X = dataset['X']
y = dataset['y']
y = dataset['y']
Xtrain, Xtest, ytrain, ytest = train_test_split(X, y, random_state=0)
# standard logistic regression
lr = logRegL2(lammy=1)
lr.fit(Xtrain, ytrain)
print("Training error %.3f" % np.mean(lr.predict(Xtrain) != ytrain))
print("Validation error %.3f" % np.mean(lr.predict(Xtest) != ytest))
utils.plotClassifier(lr, Xtrain, ytrain)
utils.savefig("logReg.png")
# kernel logistic regression with a linear kernel
lr_kernel = kernelLogRegL2(kernel_fun=logReg.kernel_linear, lammy=1)
lr_kernel.fit(Xtrain, ytrain)
print("Training error %.3f" %
np.mean(lr_kernel.predict(Xtrain) != ytrain))
print("Validation error %.3f" %
np.mean(lr_kernel.predict(Xtest) != ytest))
utils.plotClassifier(lr_kernel, Xtrain, ytrain)
utils.savefig("logRegLinearKernel.png")
elif question == "1.1":
dataset = load_dataset('nonLinearData.pkl')
X = dataset['X']
y = dataset['y']
y = dataset['y']
Xtrain, Xtest, ytrain, ytest = train_test_split(X, y, random_state=0)
# standard logistic regression
lr = logRegL2(lammy=1)
lr.fit(Xtrain, ytrain)
print("Training error %.3f" % np.mean(lr.predict(Xtrain) != ytrain))
print("Validation error %.3f" % np.mean(lr.predict(Xtest) != ytest))
utils.plotClassifier(lr, Xtrain, ytrain)
utils.savefig("logReg.png")
# kernel logistic regression with a linear kernel
lr_kernel = kernelLogRegL2(kernel_fun=logReg.kernel_linear, lammy=1)
lr_kernel.fit(Xtrain, ytrain)
print("Training error %.3f" %
np.mean(lr_kernel.predict(Xtrain) != ytrain))
print("Validation error %.3f" %
np.mean(lr_kernel.predict(Xtest) != ytest))
utils.plotClassifier(lr_kernel, Xtrain, ytrain)
utils.savefig("logRegLinearKernel.png")
elif question == "1.1":
dataset = load_dataset('nonLinearData.pkl')
X = dataset['X']
y = dataset['y']
y = dataset["y"]
Xtrain, Xtest, ytrain, ytest = train_test_split(X, y, random_state=0)
# standard logistic regression
lr = logRegL2(lammy=1)
lr.fit(Xtrain, ytrain)
print("Training error %.3f" % np.mean(lr.predict(Xtrain) != ytrain))
print("Validation error %.3f" % np.mean(lr.predict(Xtest) != ytest))
utils.plotClassifier(lr, Xtrain, ytrain)
utils.savefig("logReg.png")
# kernel logistic regression with a linear kernel
lr_kernel = kernelLogRegL2(kernel_fun=logReg.kernel_linear, lammy=1)
lr_kernel.fit(Xtrain, ytrain)
print("Training error %.3f" %
np.mean(lr_kernel.predict(Xtrain) != ytrain))
print("Validation error %.3f" %
np.mean(lr_kernel.predict(Xtest) != ytest))
utils.plotClassifier(lr_kernel, Xtrain, ytrain)
utils.savefig("logRegLinearKernel.png")
elif question == "1.1":
dataset = load_dataset("nonLinearData.pkl")
X = dataset["X"]
y = dataset["y"]
y = dataset['y']
Xtrain, Xtest, ytrain, ytest = train_test_split(X, y, random_state=0)
# standard logistic regression
lr = logRegL2(lammy=1)
lr.fit(Xtrain, ytrain)
print("Training error %.3f" % np.mean(lr.predict(Xtrain) != ytrain))
print("Validation error %.3f" % np.mean(lr.predict(Xtest) != ytest))
utils.plotClassifier(lr, Xtrain, ytrain)
utils.savefig("logReg.png")
# kernel logistic regression with a linear kernel
lr_kernel = kernelLogRegL2(kernel_fun=logReg.kernel_linear, lammy=1)
lr_kernel.fit(Xtrain, ytrain)
print("Training error %.3f" %
np.mean(lr_kernel.predict(Xtrain) != ytrain))
print("Validation error %.3f" %
np.mean(lr_kernel.predict(Xtest) != ytest))
utils.plotClassifier(lr_kernel, Xtrain, ytrain)
utils.savefig("logRegLinearKernel.png")
elif question == "1.1":
dataset = load_dataset('nonLinearData.pkl')
X = dataset['X']
y = dataset['y']
