def generate(self, weather_source, datetimes, daily_noise_dist=None):
"""Returns a ConsumptionData instance given a particular weather
source and a list of datetimes.
Parameters
----------
weather_source : eemeter.weather.WeatherSourceBase
Weather source from which to draw outdoor temperature data.
datetimes : list of datetime objects
Periods over which to simulate consumption.
daily_noise_dist : scipy.stats.rv_continuous, default None
Noise to add to each day in a period. Noise is additive and sampled
independently for each period. e.g. scipy.stats.normal().
"""
records = [{"start": start, "end": end, "value": np.nan}
for start, end in zip(datetimes, datetimes[1:])]
cd = ConsumptionData(records, self.fuel_type,
self.consumption_unit_name, record_type="arbitrary")
periods = cd.periods()
period_daily_temps = weather_source.daily_temperatures(periods,
self.temperature_unit_name)
period_average_daily_usages = self.model.transform(period_daily_temps,self.params)
for average_daily_usage, period in zip(period_average_daily_usages,periods):
n_days = period.timedelta.days
if daily_noise_dist is not None:
average_daily_usage += np.mean(daily_noise_dist.rvs(n_days))
cd.data[period.start] = average_daily_usage * n_days
return cd
def generate_consumption_records(model, params_pre, params_post, datetimes_pre, datetimes_post, fuel_type, energy_unit, weather_source):
datetimes = datetimes_pre[:-1] + datetimes_post
records = [{"start": start, "end": end, "value": np.nan}
for start, end in zip(datetimes, datetimes[1:])]
cd = ConsumptionData(records, fuel_type, energy_unit, record_type="arbitrary")
periods = cd.periods()
periods_pre = periods[:len(datetimes_pre[:-1])]
periods_post = periods[len(datetimes_pre[:-1]):]
period_pre_daily_temps = weather_source.daily_temperatures(periods_pre, TEMPERATURE_UNIT_STR)
period_post_daily_temps = weather_source.daily_temperatures(periods_post, TEMPERATURE_UNIT_STR)
period_pre_average_daily_usages = model.transform(period_pre_daily_temps, params_pre)
period_post_average_daily_usages = model.transform(period_post_daily_temps, params_post)
daily_noise_dist = None
for average_daily_usage, period in zip(period_pre_average_daily_usages, periods_pre):
n_days = period.timedelta.days
if daily_noise_dist is not None:
average_daily_usage += np.mean(daily_noise_dist.rvs(n_days))
cd.data[period.start] = average_daily_usage * n_days
for average_daily_usage, period in zip(period_post_average_daily_usages, periods_post):
n_days = period.timedelta.days
if daily_noise_dist is not None:
average_daily_usage += np.mean(daily_noise_dist.rvs(n_days))
cd.data[period.start] = average_daily_usage * n_days
return cd
def generate_consumption_records(model, params_pre, params_post, datetimes_pre,
datetimes_post, fuel_type, energy_unit,
weather_source):
datetimes = datetimes_pre[:-1] + datetimes_post
records = [{
"start": start,
"end": end,
"value": np.nan
} for start, end in zip(datetimes, datetimes[1:])]
cd = ConsumptionData(records,
fuel_type,
energy_unit,
record_type="arbitrary")
periods = cd.periods()
periods_pre = periods[:len(datetimes_pre[:-1])]
periods_post = periods[len(datetimes_pre[:-1]):]
period_pre_daily_temps = weather_source.daily_temperatures(
periods_pre, TEMPERATURE_UNIT_STR)
period_post_daily_temps = weather_source.daily_temperatures(
periods_post, TEMPERATURE_UNIT_STR)
period_pre_average_daily_usages = model.transform(period_pre_daily_temps,
params_pre)
period_post_average_daily_usages = model.transform(period_post_daily_temps,
params_post)
daily_noise_dist = None
for average_daily_usage, period in zip(period_pre_average_daily_usages,
periods_pre):
n_days = period.timedelta.days
if daily_noise_dist is not None:
average_daily_usage += np.mean(daily_noise_dist.rvs(n_days))
cd.data[period.start] = average_daily_usage * n_days
for average_daily_usage, period in zip(period_post_average_daily_usages,
periods_post):
n_days = period.timedelta.days
if daily_noise_dist is not None:
average_daily_usage += np.mean(daily_noise_dist.rvs(n_days))
cd.data[period.start] = average_daily_usage * n_days
return cd
