# (e)
lda = LinearDiscriminantAnalysis()
lda.pred = lda.fit(x_train, y_train).predict(x_test)
print(pd.DataFrame(confusion_matrix(y_test, lda.pred), index=['y=0', 'y=1'], columns=['y_pred=0', 'y_pred=1']))
print('error rate: ', accuracy_score(y_test, lda.pred)) # 62.5%
# (f)
qda = QuadraticDiscriminantAnalysis()
qda.pred = qda.fit(x_train, y_train).predict(x_test)
print(pd.DataFrame(confusion_matrix(y_test, qda.pred), index=['y=0', 'y=1'], columns=['y_pred=0', 'y_pred=1']))
print('error rate: ', accuracy_score(y_test, qda.pred)) # 58.7%
# (g)
knn = KNeighborsClassifier(n_neighbors=1)
knn.pred = knn.fit(x_train, y_train).predict(x_test)
print('error rate: ', accuracy_score(y_test, knn.pred)) # 49%
# (h): Logistic and LDA models are the best.
# (i)
# KNN
error_rate = np.array([])
k_value = np.array([])
for i in (5, 10, 20):
knn = KNeighborsClassifier(n_neighbors=i)
knn.pred = knn.fit(x_train, y_train).predict(x_test)
k_value = np.append(k_value, i)
error_rate = np.append(error_rate, 1-accuracy_score(y_test, knn.pred))
best_k = k_value[error_rate.argmin()]
)
print("calculating DECISION TREE please waite...")
dt = DecisionTreeClassifier(random_state=0)
dt.fit(fea_train, np.array(labels_train))
dt.pred = dt.predict(fea_test)
calculate_result(np.array(labels_test), dt.pred)
print(
"------------------------------------------------------------------------"
)
print("calculating KNN please waite...")
knn = KNeighborsClassifier(n_neighbors=5)
knn.fit(fea_train, np.array(labels_train))
knn.pred = knn.predict(fea_test)
calculate_result(np.array(labels_test), knn.pred)
print(
"------------------------------------------------------------------------"
)
print("calculating Nueral networks please waite...")
paramters = L_layer_model(fea_train.T,
np.array(labels_train).T, [50, 100, 50, 10])
nnpred = NN_predict(fea_test.T, np.array(labels_test).T, paramters)
calculate_result(np.array(labels_test), np.array(nnpred))
print(
"========================================================================="
)
print("Test optimal model on test set")
plt.ylabel('Error Rate')
print("the misclassification error for each k value is : ",
np.round(MSE, 3))
return nearest_k
query = list(range(0, 50))
optimal_k = find_optimal_k(X_train, Y_train, query)
# K-NN with optimal K
knn = KNeighborsClassifier(n_neighbors=optimal_k)
knn.fit(X_train, Y_train)
pred = knn.pred(X_test)
"""
Model Evaluation
"""
# Confusion Matrix
plt.figure()
cm = confusion_matrix(Y_test, pred)
class_label = ["negative", "positive"]
df_cm_test = pd.DataFrame(cm, index=class_label, columns=class_label)
sns.heatmap(df_cm_test, annot=True, fmt="d")
plt.title("Confusion Matrix for Test datas")
plt.xlabel("Predicted Label")
plt.ylabel("True Label")
# classification report
