# 학습 데이터 분리
features_training_data = [
'open_lastclose_ratio', 'high_close_ratio', 'low_close_ratio',
'close_lastclose_ratio', 'volume_lastvolume_ratio', 'close_ma5_ratio',
'volume_ma5_ratio', 'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio', 'close_ma60_ratio',
'volume_ma60_ratio', 'close_ma120_ratio', 'volume_ma120_ratio'
]
training_data = training_data[features_training_data]
# 강화학습 시작
policy_learner = PolicyLearner(stock_code=stock_code,
chart_data=chart_data,
training_data=training_data,
min_trading_unit=1,
max_trading_unit=2,
delayed_reward_threshold=.2,
lr=.001)
policy_learner.fit(balance=10000000,
num_epoches=1000,
discount_factor=0,
start_epsilon=.5)
# 정책 신경망을 파일로 저장
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % stock_code)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
model_path = os.path.join(model_dir, 'model_%s.h5' % timestr)
policy_learner.policy_network.save_model(model_path)
chart_data=chart_data,
training_data=training_data,
policy_model_path=os.path.join(
settings.BASE_DIR,
'models/{}/model_{}.h5'.format(model_code, policy_model_ver)),
value_model_path=os.path.join(
settings.BASE_DIR,
'models/{}/model_{}.h5'.format(model_code, value_model_ver)),
lr=0.00000001,
discount_factor=0,
start_epsilon=0,
num_past_input=119,
load_weight_and_learn=False)
# 여기가 핵심
policy_learner.fit(balance=1, num_epoches=1000)
# 정책 신경망을 파일로 저장
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % model_code)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
model_path = os.path.join(model_dir, 'model_policy_%s.h5' % timestr)
policy_learner.policy_network.save(model_path,
include_optimizer=False,
overwrite=True)
#policy_learner.policy_network_obj.save_model(model_path)
model_path = os.path.join(model_dir, 'model_value_%s.h5' % timestr)
policy_learner.value_network.save(model_path,
include_optimizer=False,
overwrite=True)
#policy_learner.value_network_obj.save_weights(model_path)
(training_data['date'] <= '2018-12-31')]
training_data = training_data.dropna()
# Chart Data Separation
features_chart_data = ['date', 'open', 'high', 'low', 'close', 'volume']
chart_data = training_data[features_chart_data]
# Training data separation
features_training_data = [
'open_lastclose_ratio', 'high_close_ratio', 'low_close_ratio',
'close_lastclose_ratio', 'volume_lastvolume_ratio',
'close_ma5_ratio', 'volume_ma5_ratio',
'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio'
]
training_data = training_data[features_training_data]
# Strat reinforcement learning
policy_learner = PolicyLearner(
stock_code=stock_code, chart_data=chart_data, training_data=training_data,
min_trading_unit=1, max_trading_unit=2, delayed_reward_threshold=reward, lr=.0001,tax=tax)
policy_learner.fit(balance=bal, num_epoches=500,
discount_factor=0, start_epsilon=.5,monkey=monkey)
# Save Policy Neural Network to File
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % stock_code)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
model_path = os.path.join(model_dir, 'model_%s.h5' % timestr)
policy_learner.policy_network.save_model(model_path)
stream_handler = logging.StreamHandler() # send logging output to stdout
file_handler.setLevel(logging.DEBUG)
stream_handler.setLevel(logging.INFO)
logging.basicConfig(format="%(message)s",
handlers=[file_handler, stream_handler], level=logging.DEBUG)
# 강화학습 시작
if LEARNING:
chart_data, data = prepare_data(STOCK_CODE, MARKET_CODE, TRAINING_START_DATE, TRAINING_END_DATE)
policy_learner = PolicyLearner(
stock_code=STOCK_CODE, chart_data=chart_data, training_data=data,
min_trading_unit=MIN_TRADING_UNIT, max_trading_unit=MAX_TRADING_UNIT,
delayed_reward_threshold=DELAYED_REWARD_THRESHOLD, lr=LEARNING_RATE)
policy_learner.fit(balance=INITIAL_BALANCE, num_epoches=NUM_EPOCHS,max_memory=MAX_MEMORY,
discount_factor=DISCOUNT_FACTOR, start_epsilon=START_EPSILON, learning=LEARNING_RATE)
# 정책 신경망을 파일로 저장
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % STOCK_CODE)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
#model_path = os.path.join(model_dir, 'model_%s.h5' % timestr)
model_path = os.path.join(model_dir, 'model_%s.h5' % STOCK_CODE)
policy_learner.policy_network.save_model(model_path)
if SIMULATION:
chart_data, data = prepare_data(STOCK_CODE, MARKET_CODE, SIMULATION_START_DATE, SIMULATION_END_DATE)
policy_learner = PolicyLearner(
stock_code=STOCK_CODE, chart_data=chart_data, training_data=data,
min_trading_unit=MIN_TRADING_UNIT, max_trading_unit=MAX_TRADING_UNIT)
def learnFunc(self):
if self.code is None or self.df is None:
return
self.change_value.emit(ZERO)
# 데이터 전처리
code = self.code
chart_data = self.df
prep_data = data_manager.preprocess(chart_data)
training_data = data_manager.build_training_data(prep_data)
training_data = training_data.dropna()
# 차트데이터 분리
feature_chart_data = ['date', 'open', 'high', 'low', 'close', 'volume']
chart_data = training_data[feature_chart_data]
# emit
self.change_value.emit(TWENTY_FIVE)
# 학습데이터 분리
feature_chart_data = [
'open_lastclose_ratio',
'high_close_ratio',
'low_close_ratio',
'close_lastclose_ratio',
'volume_lastvolume_ratio',
'close_ma5_ratio',
'volume_ma5_ratio',
'close_ma10_ratio',
'volume_ma10_ratio',
'close_ma20_ratio',
'volume_ma20_ratio',
'close_ma60_ratio',
'volume_ma60_ratio',
'close_ma120_ratio',
'volume_ma120_ratio',
]
training_data = training_data[feature_chart_data]
# 정책 신경망을 파일로 저장
self.createFolder('model')
mdir = os.path.join(settings.BASE_DIR, 'model')
self.createFolder(os.path.join(mdir, code))
model_dir = os.path.join(mdir, code)
model_path = os.path.join(model_dir, 'model%s.h5' % code)
# model_path 경로가 없으면 학습모델을 해당 dir에 만들어서 학습
# model_path가 있으면 해당 모델 선택 후 예측
print(model_path)
# emit
self.change_value.emit(FIFTY)
if not os.path.isfile(model_path):
start_time = time.time()
policy_learner = PolicyLearner(stock_code=code,
chart_data=chart_data,
training_data=training_data,
fig=self.fig,
canvas=self.canvas,
min_trading_unit=1,
max_trading_unit=2,
delayed_reward_threshold=0.2,
lr=0.001)
policy_learner.fit(balance=10000000,
num_epoches=200,
discount_factor=0,
start_epsilon=0.5)
end_time = time.time()
policy_learner.policy_network.save_model(model_path)
print("LearningTime: {} sec".format(end_time - start_time))
else:
start_time = time.time()
policy_learner = PolicyLearner(stock_code=code,
chart_data=chart_data,
training_data=training_data,
fig=self.fig,
canvas=self.canvas,
min_trading_unit=1,
max_trading_unit=2)
end_time = time.time()
print("LearningTime: {} sec".format(end_time - start_time))
policy_learner.trade(balance=1000000,
model_path=os.path.join(
model_dir, 'model%s.h5' % (code)))
# emit
self.change_value.emit(A_HUNDRED)
# 학습 데이터 분리
features_training_data = [
'open_lastclose_ratio', 'high_close_ratio', 'low_close_ratio',
'close_lastclose_ratio', 'volume_lastvolume_ratio', 'close_ma5_ratio',
'volume_ma5_ratio', 'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio', 'close_ma60_ratio',
'volume_ma60_ratio', 'close_ma120_ratio', 'volume_ma120_ratio'
]
training_data = training_data[features_training_data]
# 강화학습 시작
policy_learner = PolicyLearner(stock_code=stock_code,
chart_data=chart_data,
training_data=training_data,
min_trading_unit=1,
max_trading_unit=2,
delayed_reward_threshold=.2,
lr=.001)
policy_learner.fit(balance=10000000,
num_epoches=100,
discount_factor=0,
start_epsilon=.5,
index_change_rate=index_change_rate)
# 정책 신경망을 파일로 저장
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % stock_code)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
model_path = os.path.join(model_dir, 'model_%s.h5' % timestr)
policy_learner.policy_network.save_model(model_path)
