`sustaingym.envs.evcharging.utils`#

Implements utility methods for interacting with ACN-data and GMMs.

When run on its own, this script downloads the default data found in: sustaingym/data/evcharging/acn_data.

Module Contents#

Functions#

`to_la_dt`(→ datetime.datetime)	Converts string '%Y-%m-%d' to datetime localized in LA Time.
`site_str_to_site`(→ acnportal.acnsim.ChargingNetwork)	Returns charging network from string.
`get_sessions`(→ collections.abc.Iterator[dict[str, Any]])	Retrieves charging sessions using ACNData.
`fetch_real_events`(→ pandas.DataFrame)	Returns a pandas DataFrame of charging events from ACN-Data.
`get_real_events`(→ pandas.DataFrame)	Returns a pandas DataFrame of charging events.
`get_model_name`(→ str)	Returns folder name for a trained GMM.
`save_gmm_model`(→ None)	Saves GMM (presumably trained) and other information to directory.
`load_gmm_model`(→ dict[str, ...)	Load pickled GMM and other data from folder.
`round`(→ numpy.ndarray)	Round array values when decimal is above threshold.
`download_default_acndata`(→ None)	Downloads default data from ACNData.

Attributes#

`API_TOKEN`
`GMMS_DIR`
`AM_LA`
`GMT`
`DATE_FORMAT`
`DT_STRING_FORMAT`
`MINS_IN_DAY`
`ONE_DAY`
`REQ_ENERGY_SCALE`
`START_DATE`
`END_DATE`
`DEFAULT_DATE_RANGES`
`DEFAULT_PERIOD_TO_RANGE`
`DefaultPeriodStr`
`SiteStr`
`GMM_KEY`
`COUNT_KEY`
`STATION_USAGE_KEY`

sustaingym.envs.evcharging.utils.API_TOKEN = 'DEMO_TOKEN'[source]#

sustaingym.envs.evcharging.utils.GMMS_DIR = 'gmms'[source]#

sustaingym.envs.evcharging.utils.AM_LA[source]#

sustaingym.envs.evcharging.utils.GMT[source]#

sustaingym.envs.evcharging.utils.DATE_FORMAT = '%Y-%m-%d'[source]#

sustaingym.envs.evcharging.utils.DT_STRING_FORMAT = '%a, %d %b %Y %H:%M:%S GMT'[source]#

sustaingym.envs.evcharging.utils.MINS_IN_DAY = 1440[source]#

sustaingym.envs.evcharging.utils.ONE_DAY[source]#

sustaingym.envs.evcharging.utils.REQ_ENERGY_SCALE = 100[source]#

sustaingym.envs.evcharging.utils.START_DATE[source]#

sustaingym.envs.evcharging.utils.END_DATE[source]#

sustaingym.envs.evcharging.utils.DEFAULT_DATE_RANGES = (('2019-05-01', '2019-08-31'), ('2019-09-01', '2019-12-31'), ('2020-02-01', '2020-05-31'),...[source]#

sustaingym.envs.evcharging.utils.DEFAULT_PERIOD_TO_RANGE[source]#

sustaingym.envs.evcharging.utils.DefaultPeriodStr[source]#

sustaingym.envs.evcharging.utils.SiteStr[source]#

sustaingym.envs.evcharging.utils.GMM_KEY = 'gmm'[source]#

sustaingym.envs.evcharging.utils.COUNT_KEY = 'count'[source]#

sustaingym.envs.evcharging.utils.STATION_USAGE_KEY = 'station_usage'[source]#

sustaingym.envs.evcharging.utils.to_la_dt(s: str) → datetime.datetime[source]#

Converts string ‘%Y-%m-%d’ to datetime localized in LA Time.

Parameters:: s (str) –
Return type:: datetime.datetime

sustaingym.envs.evcharging.utils.site_str_to_site(site: SiteStr) → acnportal.acnsim.ChargingNetwork[source]#

Returns charging network from string.

Parameters:: site (SiteStr) –
Return type:: acnportal.acnsim.ChargingNetwork

sustaingym.envs.evcharging.utils.get_sessions(start_date: datetime.datetime, end_date: datetime.datetime, site: SiteStr = 'caltech') → collections.abc.Iterator[dict[str, Any]][source]#

Retrieves charging sessions using ACNData.

Parameters:

start_date (datetime.datetime) – beginning time of interval. Only year, month, and day are considered. The datetime is expected to be localized in LA time, the timezone of the charging garages.
end_date (datetime.datetime) – ending time of interval, exclusive. See start_date.
site (SiteStr) – ‘caltech’ or ‘jpl’

Returns:

sessions – iterator of sessions with a connection time starting on start_date and ending the day before end_date

Return type:

collections.abc.Iterator[dict[str, Any]]

Example:

fall2020_sessions = get_sessions(
        datetime(2020, 9, 1), datetime(2020, 12, 1))

sustaingym.envs.evcharging.utils.fetch_real_events(start_date: datetime.datetime, end_date: datetime.datetime, site: SiteStr) → pandas.DataFrame[source]#

Returns a pandas DataFrame of charging events from ACN-Data.

See get_sessions() for arguments.

Returns:

events – DataFrame containing charging info

arrival                   datetime64[ns, America/Los_Angeles]
departure                 datetime64[ns, America/Los_Angeles]
requested_energy (kWh)    float64
delivered_energy (kWh)    float64
station_id                str
session_id                str
estimated_departure       datetime64[ns, America/Los_Angeles]
claimed                   bool

Parameters:

start_date (datetime.datetime) –
end_date (datetime.datetime) –
site (SiteStr) –

Return type:

pandas.DataFrame

sustaingym.envs.evcharging.utils.get_real_events(start_date: datetime.datetime, end_date: datetime.datetime, site: SiteStr) → pandas.DataFrame[source]#

Returns a pandas DataFrame of charging events.

Either loads data from package or retrieves from ACN-Data.

See fetch_real_events() for arguments and return value, except function is now inclusive of end_date.

Parameters:

start_date (datetime.datetime) –
end_date (datetime.datetime) –
site (SiteStr) –

Return type:

pandas.DataFrame

sustaingym.envs.evcharging.utils.get_model_name(begin: datetime.datetime, end: datetime.datetime, n_components: int) → str[source]#

Returns folder name for a trained GMM.

Parameters:

begin (datetime.datetime) –
end (datetime.datetime) –
n_components (int) –

Return type:

str

sustaingym.envs.evcharging.utils.save_gmm_model(site: SiteStr, gmm: sklearn.mixture.GaussianMixture, cnt: numpy.ndarray, sid: numpy.ndarray, begin: datetime.datetime, end: datetime.datetime, n_components: int) → None[source]#

Saves GMM (presumably trained) and other information to directory.

Parameters:

site (SiteStr) – either ‘caltech’ or ‘jpl’
gmm (sklearn.mixture.GaussianMixture) – trained Gaussian Mixture Model
cnt (numpy.ndarray) – a 1-D np.ndarray session counts per day during date period, expected to have the same length as the number of days, inclusive, in the date period
sid (numpy.ndarray) – a 1-D np.ndarray stations’ usage counts for entire date period, expected to have the same length as the number of stations in the network
begin (datetime.datetime) – beginning of training period, for folder name
end (datetime.datetime) – ending of training period, for folder name
n_components (int) – number of GMM components

Return type:

None

sustaingym.envs.evcharging.utils.load_gmm_model(site: SiteStr, begin: datetime.datetime, end: datetime.datetime, n_components: int) → dict[str, numpy.ndarray | sklearn.mixture.GaussianMixture][source]#

Load pickled GMM and other data from folder.

If searching for a custom model, searches relative to the current working directory in GMMS_DIR. If searching for a default model, searches inside the data folder.

Parameters:

site (SiteStr) – either ‘caltech’ or ‘jpl’
begin (datetime.datetime) – start date of date range GMM is trained in
end (datetime.datetime) – end date of date range GMM is trained in
n_components (int) – number of GMM components

Returns:

data – dict containing the following key-value pairs:

‘gmm’: mixture.GaussianMixture, trained gmm, date range and components are specified on folder
‘count’: np.ndarray, session counts per day
‘station_usage’: np.ndarray, stations’ usage counts for date range

Return type:

dict[str, numpy.ndarray | sklearn.mixture.GaussianMixture]

sustaingym.envs.evcharging.utils.round(arr: numpy.ndarray, thresh: float = 0.7) → numpy.ndarray[source]#

Round array values when decimal is above threshold.

Same as np.round if thresh = 0.5

Parameters:

arr (numpy.ndarray) – input array
thresh (float) – decimal between 0 and 1

Returns:

rounded array

Return type:

numpy.ndarray

sustaingym.envs.evcharging.utils.download_default_acndata() → None[source]#

Downloads default data from ACNData.

Return type:: None

sustaingym.envs.evcharging.utils#

Module Contents#

Functions#

Attributes#

`sustaingym.envs.evcharging.utils`#