Coverage for source/training/training_config.py: 98%
42 statements
« prev ^ index » next coverage.py v7.8.0, created at 2025-07-30 20:59 +0000
1# training/training_config.py
3# global imports
4import pandas as pd
5from typing import Any, Optional
7# local imports
8from source.agent import AgentHandler, LearningStrategyHandlerBase, TestingStrategyHandlerBase
9from source.environment import LabelAnnotatorBase, LabeledDataBalancer, PriceRewardValidator, \
10 RewardValidatorBase, SimpleLabelAnnotator, TradingEnvironment
11from source.model import BluePrintBase
13class TrainingConfig():
14 """
15 Implements a configuration class for training agents in a trading environment. It encapsulates
16 all necessary parameters for training, including the number of steps, episodes, model blueprint,
17 data path, initial budget, maximum amount of trades, window size, and various reward parameters.
18 It also provides methods for instantiating an agent handler and printing the configuration.
19 """
21 def __init__(self, nr_of_steps: int, nr_of_episodes: int, model_blue_print: BluePrintBase,
22 data: pd.DataFrame, initial_budget: float, max_amount_of_trades: int, window_size: int,
23 learning_strategy_handler: LearningStrategyHandlerBase,
24 testing_strategy_handler: TestingStrategyHandlerBase, sell_stop_loss: float = 0.8,
25 sell_take_profit: float = 1.2, buy_stop_loss: float = 0.8, buy_take_profit: float = 1.2,
26 penalty_starts: int = 0, penalty_stops: int = 10, static_reward_adjustment: float = 1,
27 repeat_test: int = 10, test_ratio: float = 0.2, validator: Optional[RewardValidatorBase] = None,
28 label_annotator: Optional[LabelAnnotatorBase] = None,
29 labeled_data_balancer: Optional[LabeledDataBalancer] = None,
30 meta_data: Optional[dict[str, Any]] = None) -> None:
31 """
32 Class constructor. Initializes the training configuration with the provided parameters.
34 Parameters:
35 nr_of_steps (int): The number of training steps to perform.
36 nr_of_episodes (int): The number of training episodes to perform.
37 model_blue_print (BluePrintBase): The blueprint for the model to be trained.
38 data (pd.DataFrame): The training data.
39 initial_budget (float): The initial budget for the trading agent.
40 max_amount_of_trades (int): The maximum number of trades to perform.
41 window_size (int): The size of the observation window.
42 learning_strategy_handler (LearningStrategyHandlerBase): The handler for the learning strategy.
43 testing_strategy_handler (TestingStrategyHandlerBase): The handler for the testing strategy.
44 sell_stop_loss (float): The stop loss threshold for selling.
45 sell_take_profit (float): The take profit threshold for selling.
46 buy_stop_loss (float): The stop loss threshold for buying.
47 buy_take_profit (float): The take profit threshold for buying.
48 penalty_starts (int): The step at which to start applying penalties.
49 penalty_stops (int): The step at which to stop applying penalties.
50 static_reward_adjustment (float): The static adjustment factor for rewards.
51 repeat_test (int): The number of times to repeat testing.
52 test_ratio (float): The ratio of data to use for testing.
53 validator (Optional[RewardValidatorBase]): The reward validator to use. Defaults to PriceRewardValidator.
54 label_annotator (Optional[LabelAnnotatorBase]): The label annotator to use. Defaults to SimpleLabelAnnotator.
55 labeled_data_balancer (Optional[LabeledDataBalancer]): The labeled data balancer to use. Defaults to None.
56 meta_data (Optional[dict[str, Any]]): Optional metadata for the training configuration.
57 """
59 if validator is None:
60 validator = PriceRewardValidator()
62 if label_annotator is None:
63 label_annotator = SimpleLabelAnnotator()
65 # Training config
66 self.nr_of_steps: int = nr_of_steps
67 self.nr_of_episodes: int = nr_of_episodes
68 self.repeat_test: int = repeat_test
70 # Environment config
71 self.__data: pd.DataFrame = data
72 self.__meta_data: Optional[dict[str, Any]] = meta_data
73 self.__test_ratio: float = test_ratio
74 self.__initial_budget: float = initial_budget
75 self.__max_amount_of_trades: int = max_amount_of_trades
76 self.__window_size: int = window_size
77 self.__sell_stop_loss: float = sell_stop_loss
78 self.__sell_take_profit: float = sell_take_profit
79 self.__buy_stop_loss: float = buy_stop_loss
80 self.__buy_take_profit: float = buy_take_profit
81 self.__penalty_starts: int = penalty_starts
82 self.__penalty_stops: int = penalty_stops
83 self.__static_reward_adjustment: float = static_reward_adjustment
84 self.__validator: RewardValidatorBase = validator
85 self.__label_annotator: LabelAnnotatorBase = label_annotator
86 self.__labeled_data_balancer: Optional[LabeledDataBalancer] = labeled_data_balancer
88 # Agent config
89 self.__model_blue_print: BluePrintBase = model_blue_print
90 self.__learning_strategy_handler: LearningStrategyHandlerBase = learning_strategy_handler
91 self.__testing_strategy_handler: TestingStrategyHandlerBase = testing_strategy_handler
93 def __str__(self) -> str:
94 """
95 Returns a string representation of the configuration.
97 Creates a formatted multi-line string containing all configuration
98 parameters and their values for easy logging.
100 Returns:
101 str: Formatted string representation of the configuration.
102 """
104 labeled_data_balancer_info = ""
105 if self.__labeled_data_balancer is not None:
106 labeled_data_balancer_info = \
107 f"\tlabeled_data_balancer: {self.__labeled_data_balancer.__class__.__name__}\n" \
108 f"\t\t{vars(self.__labeled_data_balancer)}\n"
109 else:
110 labeled_data_balancer_info = "\tlabeled_data_balancer: None\n"
112 return f"Training config:\n" \
113 f"\tnr_of_steps: {self.nr_of_steps}\n" \
114 f"\tnr_of_episodes: {self.nr_of_episodes}\n" \
115 f"\trepeat_test: {self.repeat_test}\n" \
116 f"\ttest_ratio: {self.__test_ratio}\n" \
117 f"\tinitial_budget: {self.__initial_budget}\n" \
118 f"\tmax_amount_of_trades: {self.__max_amount_of_trades}\n" \
119 f"\twindow_size: {self.__window_size}\n" \
120 f"\tsell_stop_loss: {self.__sell_stop_loss}\n" \
121 f"\tsell_take_profit: {self.__sell_take_profit}\n" \
122 f"\tbuy_stop_loss: {self.__buy_stop_loss}\n" \
123 f"\tbuy_take_profit: {self.__buy_take_profit}\n" \
124 f"\tpenalty_starts: {self.__penalty_starts}\n" \
125 f"\tpenalty_stops: {self.__penalty_stops}\n" \
126 f"\tstatic_reward_adjustment: {self.__static_reward_adjustment}\n" \
127 f"\tvalidator: {self.__validator.__class__.__name__}\n" \
128 f"\t\t{vars(self.__validator)}\n" \
129 f"\tlabel_annotator: {self.__label_annotator.__class__.__name__}\n" \
130 f"\t\t{vars(self.__label_annotator)}\n" \
131 f"{labeled_data_balancer_info}" \
132 f"\tmodel_blue_print: {self.__model_blue_print.__class__.__name__}\n" \
133 f"\t\t{vars(self.__model_blue_print)}\n" \
134 f"\tlearning_strategy_handler: {self.__learning_strategy_handler.__class__.__name__}\n" \
135 f"\t\t{vars(self.__learning_strategy_handler)}\n" \
136 f"\ttesting_strategy_handler: {self.__testing_strategy_handler.__class__.__name__}\n" \
137 f"\t\t{vars(self.__testing_strategy_handler)}\n"
139 def instantiate_agent_handler(self) -> AgentHandler:
140 """
141 Instantiates the agent handler with the configured environment and strategies.
143 Returns:
144 (AgentHandler): An instance of the agent handler configured with the model blueprint,
145 trading environment, learning strategy handler and testing strategy handler.
146 """
148 environment = TradingEnvironment(self.__data, self.__initial_budget, self.__max_amount_of_trades,
149 self.__window_size, self.__validator, self.__label_annotator,
150 self.__sell_stop_loss, self.__sell_take_profit, self.__buy_stop_loss,
151 self.__buy_take_profit, self.__test_ratio, self.__penalty_starts,
152 self.__penalty_stops, self.__static_reward_adjustment,
153 self.__labeled_data_balancer, self.__meta_data)
155 return AgentHandler(self.__model_blue_print, environment, self.__learning_strategy_handler,
156 self.__testing_strategy_handler)