# Fit One-vs-all Logistic Regression
model = linear_model.logLinearClassifier(maxEvals=500, verbose=0)
model.fit(XMulti, yMulti)
print("logLinearClassifier Training error %.3f" % utils.classification_error(model.predict(XMulti), yMulti))
print("logLinearClassifier Validation error %.3f" % utils.classification_error(model.predict(XMultiValid), yMultiValid))
elif question == "3.3":
# Load Binary and Multi -class data
data = utils.load_dataset("multiData")
XMulti, yMulti = data['X'], data['y']
XMultiValid, yMultiValid = data['Xvalid'], data['yvalid']
# Fit logRegL2 model
model = linear_model.softmaxClassifier(maxEvals=500)
model.fit(XMulti, yMulti)
print("Training error %.3f" % utils.classification_error(model.predict(XMulti), yMulti))
print("Validation error %.3f" % utils.classification_error(model.predict(XMultiValid), yMultiValid))
elif question == "3.4":
data = utils.load_dataset("multiData")
XMulti, yMulti = data['X'], data['y']
XMultiValid, yMultiValid = data['Xvalid'], data['yvalid']
ovr = LogisticRegression(C=1e8, fit_intercept=False)
sft = LogisticRegression(multi_class="multinomial", solver="lbfgs", C=1e8, fit_intercept=False)
ovr.fit(XMulti, yMulti)
print("Training error %.3f" % utils.classification_error(ovr.predict(XMulti), yMulti))
model = linear_model.logLinearClassifier(maxEvals=500, verbose=0)
model.fit(XMulti, yMulti)
print("logLinearClassifier Training error %.3f" %
utils.classification_error(model.predict(XMulti), yMulti))
print("logLinearClassifier Validation error %.3f" %
utils.classification_error(model.predict(XMultiValid),
yMultiValid))
elif question == "3.3":
data = utils.load_dataset("multiData")
XMulti, yMulti = data['X'], data['y']
XMultiValid, yMultiValid = data['Xvalid'], data['yvalid']
model = linear_model.softmaxClassifier(maxEvals=500, verbose=0)
model.fit(XMulti, yMulti)
print("Training error %.3f" %
utils.classification_error(model.predict(XMulti), yMulti))
print("Validation error %.3f" % utils.classification_error(
model.predict(XMultiValid), yMultiValid))
elif question == "3.4":
data = utils.load_dataset("multiData")
XMulti, yMulti = data['X'], data['y']
XMultiValid, yMultiValid = data['Xvalid'], data['yvalid']
# One-vs-all
model = LogisticRegression(multi_class='ovr',
fit_intercept=False,
binarizer = LabelBinarizer()
Y = binarizer.fit_transform(y)
y_int = np.int32(y)
lams = [2, 1.75, 1.5, 1.25, 1]
test_error = []
train_error = []
train_bias = np.ones((X.shape[0], 1))
test_bias = np.ones((Xtest.shape[0], 1))
X = np.hstack((train_bias, X))
Xtest = np.hstack((test_bias, Xtest))
for lammy in lams:
model = linear_model.softmaxClassifier(lammy=lammy, epochs=10, alpha=1, batch=5000)
model.fit(X, y, Y)
pred = model.predict(Xtest)
e = utils.classification_error(ytest, pred)
print("at lambda ", lammy, "validation error is ", e)
test_error = np.append(test_error, e)
pred = model.predict(X)
e = utils.classification_error(y, pred)
print("at lambda ", lammy, "train error is ", e)
train_error = np.append(train_error, e)
plt.plot(lams, test_error, label="validation error")
plt.plot(lams, train_error, label="training error")
plt.title("Multi-Class Linear Classifier")
plt.xlabel("Lambda")
plt.ylabel("Error")
print("Test error = {:.4f}".format(testErr))
print('Time taken: ', time.time() - t)
elif question == "knn":
# mostly hyperparameter tuning, 10 seems to be the sweet spot
t = time.time()
model = KNN(3)
model.fit(X, y)
yhat = model.predict(Xtest)
testErr = np.mean(ytest != yhat)
print('Test error = {:.4f}'.format(testErr))
print('Time taken: ', time.time() - t)
elif question == "reg":
t = time.time()
model = softmaxClassifier(maxEvals=100)
# r = np.random.randint(5, size=np.shape(X)[0])
# Xt = X[r == 0]
# yt = y[r == 0]
# model.fit(Xt,yt)
model.fit(X, y)
yhat = model.predict(Xtest)
print(Xtest)
testErr = np.mean(ytest != yhat)
print("yhat: ", yhat)
print('Test error = {:.4f}'.format(testErr))
print('Time taken: ', time.time() - t)
