def eemeter_baseline_daily(file, temperature, install_start, install_end):
"""
This method uses linear regression to create daily load profile to
serve as the counterfactual for the period where the new measure has
been installed. The key part of this model is the temperature changepoint
determined by the heating and cooling degree days (HDD & CDD).
CalTRACK refers to a standardized model used in california to measure
energy efficiency savings from various measures (i.e. LED lightbulbs,
efficient appliances, weatherization, etc.). In this case it will be
used to model energy growth instead of savings.
"""
file = 'data/cchp_daily.csv'
daily_meter_data = eemeter.meter_data_from_csv(file, freq='daily')
# get meter data suitable for fitting a baseline model
baseline_meter_data, warnings = eemeter.get_baseline_data(
daily_meter_data, end=install_start, max_days=365)
# create a design matrix (the input to the model fitting step)
baseline_design_matrix = eemeter.create_caltrack_daily_design_matrix(
baseline_meter_data,
temperature,
)
# build a daily CalTRACK model
baseline_model = eemeter.fit_caltrack_usage_per_day_model(
baseline_design_matrix)
# get a year of reporting period data
reporting_meter_data, warnings = eemeter.get_reporting_data(
daily_meter_data, start=install_end, max_days=365)
# compute metered savings for the year of the reporting period we've selected
metered_growth_dataframe, error_bands = eemeter.metered_savings(
baseline_model,
reporting_meter_data,
temperature,
with_disaggregated=True)
# change signs for load growth
metered_growth_dataframe['metered_savings'] = metered_growth_dataframe[
'metered_savings'].apply(lambda x: x * -1)
# total metered savings
additional_load = metered_growth_dataframe.metered_savings.sum()
# metrics
metrics_raw = baseline_model.json()
r_squared_adj = metrics_raw['r_squared_adj']
cvrmse_adj = metrics_raw['avgs_metrics']['cvrmse_adj']
metrics = [r_squared_adj, cvrmse_adj, additional_load]
return metered_growth_dataframe, metrics, baseline_model
def test_json_daily():
meter_data, temperature_data, sample_metadata = (
eemeter.load_sample("il-electricity-cdd-hdd-daily"))
blackout_start_date = sample_metadata["blackout_start_date"]
blackout_end_date = sample_metadata["blackout_end_date"]
# get meter data suitable for fitting a baseline model
baseline_meter_data, warnings = eemeter.get_baseline_data(
meter_data, end=blackout_start_date, max_days=365)
# create a design matrix (the input to the model fitting step)
baseline_design_matrix = eemeter.create_caltrack_daily_design_matrix(
baseline_meter_data,
temperature_data,
)
# build a CalTRACK model
baseline_model = eemeter.fit_caltrack_usage_per_day_model(
baseline_design_matrix, )
# get a year of reporting period data
reporting_meter_data, warnings = eemeter.get_reporting_data(
meter_data, start=blackout_end_date, max_days=365)
# compute metered savings
metered_savings_dataframe, error_bands = eemeter.metered_savings(
baseline_model,
reporting_meter_data,
temperature_data,
with_disaggregated=True)
# total metered savings
total_metered_savings = metered_savings_dataframe.metered_savings.sum()
# test JSON
json_str = json.dumps(baseline_model.json())
m = eemeter.CalTRACKUsagePerDayModelResults.from_json(json.loads(json_str))
# compute metered savings from the loaded model
metered_savings_dataframe, error_bands = eemeter.metered_savings(
m, reporting_meter_data, temperature_data, with_disaggregated=True)
# total metered savings
total_metered_savings_2 = metered_savings_dataframe.metered_savings.sum()
assert total_metered_savings == total_metered_savings_2
def read_sample_data(meter_data, temperature_data, sample_metadata):
# the dates if an analysis "blackout" period during which a project was performed.
blackout_start_date = sample_metadata["blackout_start_date"]
blackout_end_date = sample_metadata["blackout_end_date"]
# get meter data suitable for fitting a baseline model
baseline_meter_data, warnings = eemeter.get_baseline_data(
meter_data, end=blackout_start_date, max_days=365)
# force alignment of weather data to the read meter data
start = meter_data.iloc[0].name.to_pydatetime()
end = meter_data.iloc[-1].name.to_pydatetime()
return {
'meter_uom_id': 'energy_kWh',
'meter_data': meter_data,
'temperature_data': temperature_data,
'sample_metadata': sample_metadata,
'blackout_start_date': blackout_start_date,
'blackout_end_date': blackout_end_date,
'baseline_meter_data': baseline_meter_data,
'start': start,
'end': end
}
def read_meter_data(meter,
blackout_start_date=None,
blackout_end_date=None,
freq=None,
start=None,
end=None,
uom=None):
# get the meter data from the meter history
logger.info('read_meter_data: freq %s', freq)
out = StringIO()
# read the meter data, returns the unit of the meter data
m_uom = meter.write_meter_data_csv(out,
columns=[{
'as_of_datetime': 'start'
}, {
'value': 'value'
}],
start=start,
end=end)
out.seek(0)
if freq not in ("hourly", "daily"):
freq = None
meter_data = eeio.meter_data_from_csv(out, freq=freq)
logger.info('read_meter_data: meter_data %s', meter_data)
# force alignment of weather data to the read meter data
start = meter_data.iloc[0].name.to_pydatetime()
end = meter_data.iloc[-1].name.to_pydatetime()
logger.info('read_meter_data: meter_data from %s to %s', start, end)
# get the temperature data from the meter linked weather stations
out = StringIO()
meter.write_weather_data_csv(out,
columns=[{
'as_of_datetime': 'dt'
}, {
'temp_f': 'tempF'
}],
start=start,
end=end)
out.seek(0)
temperature_data = eeio.temperature_data_from_csv(out, freq="hourly")
logger.info('read_meter_data: temperature_data %s', temperature_data)
# we end the model on the given blackout_start_date else end it on the last data
blm_end = blackout_start_date or end
logger.info('read_meter_data: getting baseline_meter_data ending %s',
blm_end)
# get meter data suitable for fitting a baseline model
baseline_meter_data, warnings = eemeter.get_baseline_data(meter_data,
end=blm_end,
max_days=365)
logger.info('read_meter_data: baseline_meter_data %s', baseline_meter_data)
start = baseline_meter_data.iloc[0].name.to_pydatetime()
end = baseline_meter_data.iloc[-1].name.to_pydatetime()
logger.info('read_meter_data: baseline_meter_data from %s to %s', start,
end)
logger.info('read_meter_data: DONE')
return {
'meter_uom_id': m_uom.uom_id if m_uom else None,
'meter_data': meter_data,
'temperature_data': temperature_data,
'blackout_start_date': blackout_start_date,
'blackout_end_date': blackout_end_date,
'baseline_meter_data': baseline_meter_data,
'start': start,
'end': end
}
def test_json_hourly():
meter_data, temperature_data, sample_metadata = (
eemeter.load_sample("il-electricity-cdd-hdd-hourly"))
blackout_start_date = sample_metadata["blackout_start_date"]
blackout_end_date = sample_metadata["blackout_end_date"]
# get meter data suitable for fitting a baseline model
baseline_meter_data, warnings = eemeter.get_baseline_data(
meter_data, end=blackout_start_date, max_days=365)
# create a design matrix for occupancy and segmentation
preliminary_design_matrix = (
eemeter.create_caltrack_hourly_preliminary_design_matrix(
baseline_meter_data,
temperature_data,
))
# build 12 monthly models - each step from now on operates on each segment
segmentation = eemeter.segment_time_series(preliminary_design_matrix.index,
'three_month_weighted')
# assign an occupancy status to each hour of the week (0-167)
occupancy_lookup = eemeter.estimate_hour_of_week_occupancy(
preliminary_design_matrix,
segmentation=segmentation,
)
# assign temperatures to bins
temperature_bins = eemeter.fit_temperature_bins(
preliminary_design_matrix,
segmentation=segmentation,
)
# build a design matrix for each monthly segment
segmented_design_matrices = (
eemeter.create_caltrack_hourly_segmented_design_matrices(
preliminary_design_matrix,
segmentation,
occupancy_lookup,
temperature_bins,
))
# build a CalTRACK hourly model
baseline_model = eemeter.fit_caltrack_hourly_model(
segmented_design_matrices,
occupancy_lookup,
temperature_bins,
)
# get a year of reporting period data
reporting_meter_data, warnings = eemeter.get_reporting_data(
meter_data, start=blackout_end_date, max_days=365)
# compute metered savings
metered_savings_dataframe, error_bands = eemeter.metered_savings(
baseline_model,
reporting_meter_data,
temperature_data,
with_disaggregated=True)
# total metered savings
total_metered_savings = metered_savings_dataframe.metered_savings.sum()
# test JSON
json_str = json.dumps(baseline_model.json())
m = eemeter.CalTRACKHourlyModelResults.from_json(json.loads(json_str))
# compute metered savings from the loaded model
metered_savings_dataframe, error_bands = eemeter.metered_savings(
m, reporting_meter_data, temperature_data, with_disaggregated=True)
# total metered savings
total_metered_savings_2 = metered_savings_dataframe.metered_savings.sum()
assert total_metered_savings == total_metered_savings_2
def eemeter_baseline_ami(file, temperature, install_start, install_end):
"""
This method uses linear regression to create hourly load profile to
serve as the counterfactual for the period where the new measure has
been installed. The two key parts of this model are the temperature
and occupancy binning.
CalTRACK refers to a standardized model used in california to measure
energy efficiency savings from various measures (i.e. LED lightbulbs,
efficient appliances, weatherization, etc.). In this case it will be
used to model energy growth instead of savings.
"""
ami_data = eemeter.meter_data_from_csv(file, freq='hourly')
# get meter data suitable for fitting a baseline model
baseline_meter_data, warnings = eemeter.get_baseline_data(
ami_data, end=install_start, max_days=365)
# create design matrix for occupancy and segmentation
preliminary_design_matrix = (
eemeter.create_caltrack_hourly_preliminary_design_matrix(
baseline_meter_data,
temperature,
))
# build matrix with weights for monthly models of:
# 0.5 = prior month
# 1.0 = current month
# 0.5 = post month
segmentation = eemeter.segment_time_series(
preliminary_design_matrix.index,
'three_month_weighted' # using 3 month weighted approach
)
# assign an occupancy status to each hour of the week (0-167)
occupancy_lookup = eemeter.estimate_hour_of_week_occupancy(
preliminary_design_matrix,
segmentation=segmentation,
)
# assign temperatures to bins
temperature_bins = eemeter.fit_temperature_bins(
preliminary_design_matrix,
segmentation=segmentation,
)
# build a desgin matrix for each monthly segment
segmented_design_matrices = (
eemeter.create_caltrack_hourly_segmented_design_matrices(
preliminary_design_matrix,
segmentation,
occupancy_lookup,
temperature_bins,
))
# build a CalTRACK hourly model
baseline_model = eemeter.fit_caltrack_hourly_model(
segmented_design_matrices, occupancy_lookup, temperature_bins)
# get a year of post installation reporting data
reporting_meter_data, warnings = eemeter.get_reporting_data(
ami_data, start=install_end, max_days=365)
# compute metered load growth for the year of reporitng period
metered_growth_dataframe, error_bands = eemeter.metered_savings(
baseline_model,
reporting_meter_data,
temperature,
with_disaggregated=True)
metered_growth_dataframe['temp'] = temperature # append temperature
# change signs for load growth
metered_growth_dataframe['metered_savings'] = metered_growth_dataframe[
'metered_savings'].apply(lambda x: x * -1)
# totaled load growth
additional_load = metered_growth_dataframe.metered_savings.sum()
# metrics
r_squared_adj_list = []
cvrmse_adj_list = []
# results in a dict
model_results = list(baseline_model.json().values())
results = model_results[6]
for segment, measures in results.items():
for measure, value in measures.items():
if measure == 'r_squared_adj':
r_squared_adj_list.append(value)
if measure == 'cvrmse_adj':
cvrmse_adj_list.append(value)
r_squared_adj = mean(r_squared_adj_list)
cvrmse_adj = mean(cvrmse_adj_list)
metrics = [r_squared_adj, cvrmse_adj, additional_load]
# Return Section
return metered_growth_dataframe, metrics, baseline_model
