Utils

smart_control.reinforcement_learning.utils.config

Reinforcement learning configurations.

get_histogram_path

get_histogram_path() -> str

Get path to histogram data.

Returns:

Type	Description
str	Path to histogram data.

get_histogram_reducer

get_histogram_reducer() -> Any

Get histogram reducer.

Returns:

Type	Description
Any	Histogram reducer.

get_metrics_path

get_metrics_path() -> str

Get path to metrics.

Returns:

Type	Description
str	Path to metrics.

get_reset_temp_values

get_reset_temp_values() -> np.ndarray

Get reset temperature values.

Returns:

Type	Description
ndarray	Reset temperature values.

get_weather_path

get_weather_path() -> str

Get path to weather data.

Returns:

Type	Description
str	Path to weather data.

get_zone_path

get_zone_path() -> str

Get path to zone data.

Returns:

Type	Description
str	Path to zone data.

smart_control.reinforcement_learning.utils.constants

Reinforcement learning constants.

smart_control.reinforcement_learning.utils.data_processing

Reinforcement learning data processing functions.

convert_celsius_to_kelvin

convert_celsius_to_kelvin(
    temperature_celsius: Union[float, ndarray, Series]
) -> Union[float, np.ndarray, pd.Series]

Convert temperature from Celsius to Kelvin.

Parameters:

Name	Type	Description	Default
temperature_celsius	Union[float, ndarray, Series]	Temperature in Celsius.	required

Returns:

Type	Description
Union[float, ndarray, Series]	Temperature in Kelvin.

convert_kelvin_to_celsius

convert_kelvin_to_celsius(
    temperature_kelvin: Union[float, ndarray, Series]
) -> Union[float, np.ndarray, pd.Series]

Convert temperature from Kelvin to Celsius.

Parameters:

Name	Type	Description	Default
temperature_kelvin	Union[float, ndarray, Series]	Temperature in Kelvin.	required

Returns:

Type	Description
Union[float, ndarray, Series]	Temperature in Celsius.

get_action_timeseries

get_action_timeseries(action_responses: List[Any]) -> pd.DataFrame

Converts action responses to a dataframe.

Parameters:

Name	Type	Description	Default
action_responses	List[Any]	List of action response objects.	required

Returns:

Type	Description
DataFrame	DataFrame with action timeseries data.

get_energy_timeseries

get_energy_timeseries(
    reward_infos: List[Any], time_zone: str = DEFAULT_TIME_ZONE
) -> pd.DataFrame

Returns a timeseries of energy rates.

Parameters:

Name	Type	Description	Default
reward_infos	List[Any]	List of reward info objects.	required
time_zone	str	Time zone for the timestamps.	DEFAULT_TIME_ZONE

Returns:

Type	Description
DataFrame	DataFrame with energy timeseries data.

get_latest_episode_reader

get_latest_episode_reader(metrics_path: str) -> controller_reader.ProtoReader

Get reader for the latest episode.

Parameters:

Name	Type	Description	Default
metrics_path	str	Path to metrics directory.	required

Returns:

Type	Description
ProtoReader	Reader for the latest episode.

get_outside_air_temperature_timeseries

get_outside_air_temperature_timeseries(
    observation_responses: List[Any], time_zone: str = DEFAULT_TIME_ZONE
) -> pd.Series

Returns a timeseries of outside air temperature.

Parameters:

Name	Type	Description	Default
observation_responses	List[Any]	List of observation response objects.	required
time_zone	str	Time zone for the timestamps.	DEFAULT_TIME_ZONE

Returns:

Type	Description
Series	Series with outside air temperature timeseries data.

get_reward_timeseries

get_reward_timeseries(
    reward_infos: List[Any],
    reward_responses: List[Any],
    time_zone: str = DEFAULT_TIME_ZONE,
) -> pd.DataFrame

Returns a timeseries of reward values.

Parameters:

Name	Type	Description	Default
reward_infos	List[Any]	List of reward info objects.	required
reward_responses	List[Any]	List of reward response objects.	required
time_zone	str	Time zone for the timestamps.	DEFAULT_TIME_ZONE

Returns:

Type	Description
DataFrame	DataFrame with reward timeseries data.

get_zone_timeseries

get_zone_timeseries(
    reward_infos: List[Any], time_zone: str = DEFAULT_TIME_ZONE
) -> pd.DataFrame

Converts reward infos to a timeseries dataframe.

Parameters:

Name	Type	Description	Default
reward_infos	List[Any]	List of reward info objects.	required
time_zone	str	Time zone for the timestamps.	DEFAULT_TIME_ZONE

Returns:

Type	Description
DataFrame	DataFrame with zone timeseries data.

smart_control.reinforcement_learning.utils.environment

Reinforcement learning environment.

create_and_setup_environment

create_and_setup_environment(
    gin_config_file: str,
    metrics_path: str = None,
    occupancy_normalization_constant: float = DEFAULT_OCCUPANCY_NORMALIZATION_CONSTANT,
)

Creates and sets up the environment.

load_environment

load_environment(gin_config_file: str)

Returns an Environment from a config file.

smart_control.reinforcement_learning.utils.metrics

Reinforcement learning metrics.

compute_avg_return

compute_avg_return(
    environment: Any,
    policy: PyPolicy,
    num_episodes: int = 1,
    time_zone: str = DEFAULT_TIME_ZONE,
    trajectory_observers: Optional[List[Callable]] = None,
    num_steps: int = 6,
) -> Tuple[float, List[List[Any]]]

Computes the average return of the policy on the environment.

Parameters:

Name	Type	Description	Default
environment	Any	Environment to evaluate on.	required
policy	PyPolicy	Policy to evaluate.	required
num_episodes	int	Total number of episodes to run.	1
time_zone	str	Time zone for timestamps.	DEFAULT_TIME_ZONE
trajectory_observers	Optional[List[Callable]]	List of trajectory observers.	None
num_steps	int	Number of steps to take per episode.	6

Returns:

Type	Description
float	Tuple of (average return, list of [simulation time, episode return]
List[List[Any]]	pairs).

get_trajectory

get_trajectory(
    time_step: TimeStep, current_action: PolicyStep
) -> trajectory.Trajectory

Get the trajectory for the current action and time step.

Parameters:

Name	Type	Description	Default
time_step	TimeStep	Current time step.	required
current_action	PolicyStep	Current action.	required

Returns:

Type	Description
Trajectory	Trajectory for the current action and time step.

smart_control.reinforcement_learning.utils.time_utils

Reinforcement learning time utils.

time_from_sin_cos

time_from_sin_cos(sin_theta: float, cos_theta: float) -> float

Converts sin/cos representation to radians (time angle).

to_dow

to_dow(sin_theta: float, cos_theta: float) -> int

Converts sin/cos to day of week (0-6).

to_hod

to_hod(sin_theta: float, cos_theta: float) -> int

Converts sin/cos to hour of day (0-23).