def random_forest(X,y,X_train,y_train,X_test,y_test,params):
reg = RandomForestClassifier(n_estimators = params['n_estimators'],max_depth = params['max_depth'])
reg.fit(X_train,y_train)
y_pre = reg.predict(X_test)
metrics = show_metrics('Random Forest',y_test,y_pre)
draw_roc(X,y,X_train,y_train,X_test,y_test,reg)
return metrics
def svm(X,y,X_train,y_train,X_test,y_test,params):
reg = SVC(params)
reg.fit (X_train,y_train)
y_pre = reg.predict(X_test)
metrics = show_metrics('SVM',y_test,y_pre)
draw_roc(X,y,X_train,y_train,X_test,y_test,reg)
return metrics
def logistic_regression(X,y,X_train,y_train,X_test,y_test,params):
reg = LogisticRegression(C = params)
reg.fit (X_train,y_train)
y_pre = reg.predict(X_test)
metrics = show_metrics('Logistic Regression',y_test,y_pre)
draw_roc(X,y,X_train,y_train,X_test,y_test,reg)
return metrics
length_episode[episode] += 1
total_reward_episode[episode] += reward
if is_done:
break
state = next_state
policy = {}
Q = Q1 + Q2
for state in range(n_state):
policy[state] = torch.argmax(Q[state]).item()
return Q, policy
gamma = 1
n_episode = 3000
alpha = 0.4
epsilon = 0.1
epsilon_greedy_policy = gen_epsilon_greedy_policy(env.action_space.n, epsilon)
length_episode = [0] * n_episode
total_reward_episode = [0] * n_episode
optimal_Q, optimal_policy = double_q_learning(env, gamma, n_episode, alpha)
plot_length_reward(length_episode, total_reward_episode)
show_metrics(length_episode, total_reward_episode)
print(optimal_policy)
from utils import compute_metrics, show_metrics
ROOT_PATH = '.'
#%% Train Dataframe
print("---------------------------")
print("# Train Dataset Dataframe #")
print("---------------------------")
path_to_dataset = os.path.join(ROOT_PATH, 'data', 'Full_Dataset.csv')
df = pd.read_csv(path_to_dataset)
recordid, y_C_prob, y_C_pred = testing_function(df)
y_test = df['In-hospital_death']
recall, precision, prc_auc, roc_auc = compute_metrics(y_test, y_C_prob,
y_C_pred)
show_metrics(precision, recall, prc_auc, roc_auc)
# --------------------------#
# Train Dataset Dataframe #
# --------------------------#
# Precision : 0.667
# Recall : 0.663
# Min(P,R) : 0.663
# AUPRC : 0.745
# AUROC : 0.950
#%% Test Dataframe
print("---------------------------")
print("# Test Dataset Dataframe #")
print("---------------------------")
path_to_dataset = os.path.join(ROOT_PATH, 'data', 'Testing_Dataset.csv')