Single stock prediction not working when using turbulence threshold

[rektosaure]

Hi,

Stocks models, including ensemble method, don't work with single stock in StockTradingEnv. It fails when running prediction and using turbulence_threshold: e_trade_gym = StockTradingEnv(df = trade, turbulence_threshold = 70, risk_indicator_col='vix', **env_kwargs) df_account_value, df_actions = DRLAgent.DRL_prediction( model=trained_agent, environment = e_trade_gym)

---

AttributeError Traceback (most recent call last)

in () 1 df_account_value, df_actions = DRLAgent.DRL_prediction( 2 model=trained_agent, ----> 3 environment = e_trade_gym) 3 frames /usr/local/lib/python3.7/dist-packages/finrl/neo_finrl/env_stock_trading/env_stocktrading.py in step(self, actions) 250 self.data = self.df.loc[self.day,:] 251 if self.turbulence_threshold is not None: --> 252 self.turbulence = self.data[self.risk_indicator_col].values[0] 253 self.state = self._update_state() 254 AttributeError: 'numpy.float64' object has no attribute 'values'

[Athe-kunal]

Yes, it is expecting a data frame and it is possible with more than one stock. Probably you can navigate to /usr/local/lib/python3.7/dist-packages/finrl/neo_finrl/env_stock_trading/env_stocktrading.py and substitute this if self.turbulence_threshold is not None:
if len(self.df.tic.unique()) == 1: self.turbulence = self.data[self.risk_indicator_col] elif len(self.df.tic.unique()) > 1: self.turbulence = self.data[self.risk_indicator_col].values[0] self.state = self._update_state()

[rektosaure]

Thank you