def __init__(self):
BaseAlgorithm.__init__(self)
self.get_config()
self.table_data = self.exec_sql(self.config['tableName'],
self.config['X'], self.config['Y'])
self.model = self.load_model_by_database(self.config['algorithm'],
self.config['model'])
def __init__(self):
BaseAlgorithm.__init__(self)
self.get_config()
if self.config["oneSample"]:
self.table_data = None
else:
self.table_data = self.exec_sql(self.config['tableName'], self.config['X'], None)
self.model = self.load_model_by_database(self.config['algorithm'], self.config['model'])
def __init__(self, method):
BaseAlgorithm.__init__(self)
self.one_type = "Classifier"
self.one_type_name = "分类"
self.second_type = "logisticRegression"
self.second_type_name = "逻辑回归"
# super(logisticAlgorithm, self).__init__()
if method == "train":
self.get_train_config_from_web()
elif method == "evaluate":
self.get_evaluate_config_from_web()
else:
self.get_predict_config_from_web()
if method == 'predict' and self.config["oneSample"]:
self.table_data = None
else:
self.table_data = self.exec_sql(self.config['tableName'],
self.config['X'], self.config['Y'])
def __init__(self, method):
BaseAlgorithm.__init__(self)
self.one_type = "Cluster"
self.one_type_name = "聚类"
self.second_type = "hierarchicalCluster"
self.second_type_name = "层次聚类"
# super(logisticAlgorithm, self).__init__()
if method == "train":
self.get_train_config_from_web()
elif method == "evaluate":
self.get_evaluate_config_from_web()
else:
self.get_predict_config_from_web()
if method == 'predict':
if self.config["oneSample"]:
self.table_data = None
else:
self.table_data = self.exec_sql(self.config['tableName'], self.config['X'])
else:
self.table_data = self.exec_sql(self.config['tableName'], self.config['X'])
def __init__(self, method):
BaseAlgorithm.__init__(self)
self.one_type = "Regression"
self.one_type_name = "回归"
self.second_type = "linerRegression"
self.second_type_name = "线性回归"
# super(logisticAlgorithm, self).__init__()
if method == "train":
self.get_train_config_from_web()
elif method == "evaluate":
self.get_evaluate_config_from_web()
elif method == "predict":
self.get_predict_config_from_web()
elif method == "visualization":
self.get_visualization_config_from_web()
else:
raise ValueError("input method:{} is not supported".format(method))
if method == 'predict' and self.config["oneSample"]:
self.table_data = None
else:
self.table_data = self.exec_sql(self.config['tableName'],
self.config['X'], self.config['Y'])
from dynaq_agent import DynaQAgent
from q_agent import QAgent
from random_agent import RandomAgent
repetitions = 1
single_model = True
path_to_experiment_configs = "experiments/"
default_config = 'config.json'
#algorithm = BaseAlgorithm(exploration=True, explorer=EpsilonGreedy(start=1.0, end=0.05, steps=10000),
# use_database=False, action_selection = "moving average")
algorithm = BaseAlgorithm(exploration=True,
explorer=EpsilonGreedy(start=1.0,
end=0.05,
steps=10000),
use_database=False,
action_selection="majority vote")
explorer = EpsilonGreedy(start=1., end=0.05, steps=10000)
#algorithm = QAgent(exploration=True, explorer=explorer)
#algorithm = DynaQAgent(exploration=True, explorer=explorer)
# algorithm = Bayesian_Qlearning()
#algorithm = Speedy_Qlearning(exploration=True, explorer=explorer)
#algorithm = MeanAgent(exploration=True, explorer=explorer)
#algorithm = RandomAgent()
def generate_experiments():
for lamda in np.linspace(0, 0.9, num=3, endpoint=True):
if compare_random == True:
random_scores = simulation.simulate_multiple_environment(envs, random_algorithm, T=horizon,
num_trials=num_trials,
discount=1)
mean_random = np.mean(random_scores, axis=1)
std_random = np.std(random_scores, axis=1)
for i, (score, rand_score, score_std, rand_std) in enumerate(
zip(mean_scores, mean_random, std_scores, std_random)):
print(f'Environment: "{env_names[i]}"')
print(f'-- Mean reward: {score} -- Std: {score_std}')
print(f'-- Random reward: {rand_score} -- Std: {rand_std}')
else:
for i, (score, score_std) in enumerate(zip(mean_scores, std_scores)):
print(f'Environment: "{env_names[i]}"')
print(f'-- Mean reward: {score} -- Var: {score_std}')
if save_results:
with open('results.csv', 'w') as file:
file.write('environment, runs, trials, mean_score, std_deviation\n')
for env, mean, sigma in zip(env_names, mean_scores, std_scores):
file.write('{}, {}, {}, {}, {}\n'.format(env, horizon, num_trials, mean, sigma))
return env_names, scores
if __name__ == '__main__':
algorithm = BaseAlgorithm(exploration=True, explorer=EpsilonGreedy(start=0.5, end=0.05, steps=1000),
use_database=True, action_selection = "epsilon greedy")
main(algorithm)