def test_stay(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
stay_mount = self.env.step(action='stay', mount=1)
assert stay_mount == {'success': 250000.0}
assert self.env.stock_balance == 0
assert self.env.balance == 250000
def test_buy(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
buy_mount = self.env.step(action='buy', mount=1)
assert buy_mount == {'success': 249999.987}
assert self.env.stock_balance == 1
assert self.env.balance == 249888.668
def test_reset(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
self.env.step(action='buy', mount=1)
self.env.step(action='sell', mount=1)
self.env.reset()
assert self.env.stock_balance == 0
assert self.env.balance == 250000
def test_sell(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
sell_mount = self.env.step(action='sell', mount=1)
assert sell_mount == {'fail': 0}
self.env.step(action='buy', mount=1)
sell_mount = self.env.step(action='sell', mount=1)
assert sell_mount == {'success': 249999.98500000002}
assert self.env.stock_balance == 0
assert self.env.balance == 249999.98500000002
def test_evaluate(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
self.agent = Agent()
self.agent.model.compile(optimizer=Adam(), loss="mse")
state = (self.env.balance, self.env.stock_balance)
y = self.agent.evaluate(state=state)
assert y.shape == (3, )
assert any(y) is True
def test_train(self):
env = Env(balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
agent = Agent(input_data_shape=(10, ))
mount_agent = Agent(actions=10, input_data_shape=(10, ))
print(len(env.fx_time_data_buy))
trainer = Trainer_priority(env,
agent,
mount_agent,
data_end_index=len(env.fx_time_data_buy) -
2)
trainer.train()
def test_act(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
self.agent = Agent()
self.agent.model.compile(optimizer=Adam(), loss="mse")
state = (self.env.balance, self.env.stock_balance)
action = self.agent.act(state, epsilon=0.1)
action_state = False
if action == 0 or action == 1 or action == 2:
action_state = True
assert action_state is True
def main():
env = Env()
play_pipe, predict_pipe = Pipe()
train_pipe1, train_pipe2 = Pipe()
is_training = Value("b", True)
manager = DQNManager(env.state_n, env.action_n, train_pipe1, predict_pipe,
is_training)
controller = AIControl(env, train_pipe2, play_pipe, is_training)
manager.start()
controller.control_start()
manager.join()
def main():
parser = argparse.ArgumentParser(
description='Execute train reinforcement learning.')
parser.add_argument(
'--dataset_name',
type=str,
default="../data/raw/FX_Demo/sample10000_USD_JPY_S5.pickle",
help='an integer for the accumulator')
args = parser.parse_args()
print(args.dataset_name)
env = Env(balance=250000, FX_DATA_FILE=args.dataset_name)
agent = Agent(input_data_shape=(10, ))
mount_agent = Agent(actions=10, input_data_shape=(10, ))
trainer = Trainer(env,
agent,
mount_agent,
Adam(lr=1e-6),
data_end_index=len(env.fx_time_data_buy) - 2)
trainer.train()
def main():
env = Env()
controller = AIControl(env)
controller.control_start()
class TestEnv(TestCase):
def test_sell(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
sell_mount = self.env.step(action='sell', mount=1)
assert sell_mount == {'fail': 0}
self.env.step(action='buy', mount=1)
sell_mount = self.env.step(action='sell', mount=1)
assert sell_mount == {'success': 249999.98500000002}
assert self.env.stock_balance == 0
assert self.env.balance == 249999.98500000002
def test_buy(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
buy_mount = self.env.step(action='buy', mount=1)
assert buy_mount == {'success': 249999.987}
assert self.env.stock_balance == 1
assert self.env.balance == 249888.668
def test_stay(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
stay_mount = self.env.step(action='stay', mount=1)
assert stay_mount == {'success': 250000.0}
assert self.env.stock_balance == 0
assert self.env.balance == 250000
def test_reset(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
self.env.step(action='buy', mount=1)
self.env.step(action='sell', mount=1)
self.env.reset()
assert self.env.stock_balance == 0
assert self.env.balance == 250000
else:
q_value = self.model.predict(history)
return np.argmax(q_value[0])
def load_model(self, filename):
self.model.load_weights(filename)
def pre_processing(observe):
processed_observe = np.uint8(
resize(rgb2gray(observe), (84, 84), mode='constant') * 255)
return processed_observe
if __name__ == "__main__":
env = Env()
agent = TestAgent(action_size=6, env=env)
agent.load_model("./save_model/supermario_per.h5")
for e in range(EPISODES):
done = False
max_x = 0
now_x = 0
hold_frame = 0
before_max_x = 200
start_position = 500
step, score = 0, 0
observe = env.reset(start_position=start_position)
state = pre_processing(observe)