accuracy_score(adult_tst_y, adult_pred_y)))
print("Time elapsed: {}".format(time() - start_time))
start_time = time()
dt_flag_final = tree.DecisionTreeClassifier()
dt_flag_final.fit(flag_x, flag_y)
flag_pred_y = dt_flag_final.predict(flag_tst_x)
print("Flag decision tree accuracy: {}".format(
accuracy_score(flag_tst_y, flag_pred_y)))
print("Time elapsed: {}".format(time() - start_time))
fig_adult_lc = gp.plot_learning_curve(dt_adult,
"adult learning curve",
adult_x,
adult_y,
cv=3,
train_sizes=np.linspace(0.1, 1.0, 20))
fig_adult_lc.savefig(root_path + "/plots/decision_tree/adult_lc.png")
fig_flag_lc = gp.plot_learning_curve(dt_flag,
"flag learning curve",
flag_x,
flag_y,
cv=3,
train_sizes=np.linspace(0.1, 1.0, 20))
fig_flag_lc.savefig(root_path + "/plots/decision_tree/flag_lc.png")
print("########## Plotting Max Depth Validation Curves... ##########")
fig_adult_vc1 = gp.plot_validation_curve(dt_adult,
"Max Depth Adult Validation Curve",
print('Training accuracy: ', adult_y_accuracy)
print('F1 score: ', adult_y_f1)
# Predict labels for test set and assess accuracy
adult_tst_y_pred = nn_model1.predict(adult_tst_x)
adult_tst_y_accuracy = accuracy_score(adult_tst_y, adult_tst_y_pred)
adult_tst_y_f1 = f1_score(adult_tst_y, adult_tst_y_pred, average='micro')
print('Test accuracy: ', adult_tst_y_accuracy)
print('F1 score: ', adult_tst_y_f1)
fig_adult_lc = gp.plot_learning_curve(nn_model1,
"Adult - learning curve",
adult_x,
adult_y,
cv=3,
train_sizes=np.linspace(
0.03, 1.0, 12))
fig_adult_lc.savefig(root_path + "/plots/nn/cover_lc_gd.png")
print("Time elapsed: {}".format(time() - start))
for nodes in [[10]]:
print(nodes)
start = time()
nn_model1 = mlrose.NeuralNetwork(hidden_nodes=nodes,
activation='relu',
algorithm='genetic_alg',
print("Time elapsed: {}".format(time() - start_time))
start_time = time()
nn_flag_final = neural_network.MLPClassifier(max_iter=40000,
hidden_layer_sizes=35,
alpha=0.65)
nn_flag_final.fit(flag_x.todense(), flag_y.ravel())
flag_pred_y = nn_flag_final.predict(flag_tst_x.todense())
print("Flag neural network accuracy: {}".format(
accuracy_score(flag_tst_y, flag_pred_y)))
print("Time elapsed: {}".format(time() - start_time))
fig_adult_lc = gp.plot_learning_curve(nn_adult,
"Adult - learning curve",
adult_x.todense(),
adult_y.values.ravel(),
cv=3,
train_sizes=np.linspace(0.1, 1.0, 7))
fig_adult_lc.savefig(root_path + "/plots/nn/adult_lc.png")
fig_flag_lc = gp.plot_learning_curve(nn_flag,
"Flag - learning curve",
flag_x.todense(),
flag_y.values.ravel(),
cv=3,
train_sizes=np.linspace(0.1, 1.0, 50))
fig_flag_lc.savefig(root_path + "/plots/nn/flag_lc.png")
print("########## Plotting alpha Validation Curves... ##########")
fig_adult_vc1 = gp.plot_validation_curve(nn_adult,
"Adult - alpha Validation Curve",
print("Adult SVM accuracy: {}".format(accuracy_score(adult_tst_y,
adult_pred_y)))
print("Time elapsed: {}".format(time() - start_time))
start_time = time()
svc_flag_final = svm.SVC(kernel="linear", max_iter=75000, C=0.4)
svc_flag_final.fit(flag_x.todense(), flag_y.ravel())
flag_pred_y = svc_flag_final.predict(flag_tst_x.todense())
print("Flag SVM accuracy: {}".format(accuracy_score(flag_tst_y, flag_pred_y)))
print("Time elapsed: {}".format(time() - start_time))
fig_adult_lc = gp.plot_learning_curve(svc_adult,
"Adult - learning curve",
adult_x,
adult_y.values.ravel(),
cv=5,
train_sizes=np.linspace(0.1, 1.0, 5))
fig_adult_lc.savefig(root_path + "/plots/svm/linear_adult_lc.png")
fig_flag_lc = gp.plot_learning_curve(svc_flag,
"Flag - learning curve",
flag_x.todense(),
flag_y.values.ravel(),
cv=3,
train_sizes=np.linspace(0.1, 1.0, 10))
fig_flag_lc.savefig(root_path + "/plots/svm/linear_flag_lc.png")
svc_adult2 = svm.SVC(kernel="rbf", max_iter=1000000)
svc_flag2 = svm.SVC(kernel="rbf")