def test_util_findDateIndex():
# Testing index finder for specific dates.
a = datetime.datetime(2014, 1, 1, 0, 0, 0, 0)
b = datetime.datetime(2014, 1, 2, 0, 0, 0, 0)
datetime_list = numpy.array(set_up_datetimes(a, b, 3600)).flatten()
testDateIndex = dtt.findDateIndex(
datetime_list, datetime.datetime(2014, 1, 1, 7, 0, 0, 0))
assert testDateIndex == 0
testDateIndex = dtt.findDateIndex(
datetime_list, datetime.datetime(2014, 2, 1, 0, 0, 0, 0))
assert testDateIndex == 24
def test_util_findDateIndex():
# Testing index finder for specific dates.
a = datetime.datetime(2014, 1, 1, 0, 0, 0, 0)
b = datetime.datetime(2014, 1, 2, 0, 0, 0, 0)
datetime_list = numpy.array(set_up_datetimes(a, b, 3600)).flatten()
testDateIndex = dtt.findDateIndex(
datetime_list,
datetime.datetime(2014, 1, 1, 7, 0, 0, 0)
)
assert testDateIndex == 0
testDateIndex = dtt.findDateIndex(
datetime_list,
datetime.datetime(2014, 2, 1, 0, 0, 0, 0)
)
assert testDateIndex == 24
def execute(self):
# Called after User hits GO
"""
Calculates weather sensitivity using Spearman rank.
Also, outputs data points for energy signature scatter plot.
"""
self.out.log("Starting Day Time Temperature Analysis", logging.INFO)
# Gather loads and outside air temperatures. Reduced to an hourly average
load_query = self.inp.get_query_sets('load', group_by='hour',
group_by_aggregation=Avg,
exclude={'value':None},
wrap_for_merge=True)
oat_query = self.inp.get_query_sets('oat', group_by='hour',
group_by_aggregation=Avg,
exclude={'value':None},
wrap_for_merge=True)
# Get conversion factor
base_topic = self.inp.get_topics()
meta_topics = self.inp.get_topics_meta()
load_unit = meta_topics['load'][base_topic['load'][0]]['unit']
temperature_unit = meta_topics['oat'][base_topic['oat'][0]]['unit']
load_convertfactor = cu.conversiontoKWH(load_unit)
# Match the values by timestamp
merged_load_oat = self.inp.merge(load_query, oat_query)
load_values = []
oat_values = []
datetime_values = []
for x in merged_load_oat:
if temperature_unit == 'celcius':
convertedTemp = cu.convertCelciusToFahrenheit(x['oat'][0])
elif temperature_unit == 'kelvin':
convertedTemp = cu.convertKelvinToCelcius(
cu.convertCelciusToFahrenheit(x['oat'][0]))
else:
convertedTemp = x['oat'][0]
load_values.append(x['load'][0] * load_convertfactor) #Converted to kWh
oat_values.append(convertedTemp)
datetime_values.append(dt.datetime.strptime(x['time'],'%Y-%m-%d %H'))
indexList = {}
indexList['trainingStart'] = ttow.findDateIndex(datetime_values, self.baseline_start)
self.out.log('@trainingStart '+str(indexList['trainingStart']), logging.INFO)
indexList['trainingStop'] = ttow.findDateIndex(datetime_values, self.baseline_stop)
self.out.log('@trainingStop '+str(indexList['trainingStop']), logging.INFO)
indexList['predictStart'] = ttow.findDateIndex(datetime_values, self.savings_start)
self.out.log('@predictStart '+str(indexList['predictStart']), logging.INFO)
indexList['predictStop'] = ttow.findDateIndex(datetime_values, self.savings_stop)
self.out.log('@predictStop '+str(indexList['predictStop']), logging.INFO)
for indx in indexList.keys():
if indexList[indx] == None:
self.out.log("Date not found in the datelist", logging.WARNING)
# Break up data into training and prediction periods.
timesTrain = datetime_values[indexList['trainingStart']:indexList['trainingStop']]
timesPredict = datetime_values[indexList['predictStart']:indexList['predictStop']]
valsTrain = load_values[indexList['trainingStart']:indexList['trainingStop']]
valsActual = load_values[indexList['predictStart']:indexList['predictStop']]
oatsTrain = oat_values[indexList['trainingStart']:indexList['trainingStop']]
oatsPredict = oat_values[indexList['predictStart']:indexList['predictStop']]
# Generate other information needed for model.
timeStepMinutes = (timesTrain[1] - timesTrain[0]).total_seconds()/60
# TODO: Should this be calculated in the utility function
binCt = 6 # TODO: Allow caller to pass this in as an argument.
# Form the temperature-time-of-week model.
self.out.log("Starting baseline model", logging.INFO)
ttowModel = ttow.formModel(timesTrain,
oatsTrain,
valsTrain,
timeStepMinutes,
binCt)
# Apply the model.
self.out.log("Applying baseline model", logging.INFO)
valsPredict = ttow.applyModel(ttowModel, timesPredict, oatsPredict)
# Output for scatter plot
prevSum = 0
for ctr in range(len(timesPredict)):
# Calculate cumulative savings.
prevSum += (valsPredict[ctr] - valsActual[ctr])
self.out.insert_row("DayTimeTemperatureModel", {
"datetimeValues": timesPredict[ctr],
"measured": valsActual[ctr],
"predicted": valsPredict[ctr],
"cumulativeSum": prevSum
})
def execute(self):
# Called after User hits GO
"""
Calculates weather sensitivity using Spearman rank.
Also, outputs data points for energy signature scatter plot.
"""
self.out.log("Starting application: whole building energy savings.",
logging.INFO)
# Gather loads and outside air temperatures. Reduced to an hourly average
self.out.log("Querying database.", logging.INFO)
load_query = self.inp.get_query_sets('load',
group_by='hour',
group_by_aggregation=Avg,
exclude={'value': None},
wrap_for_merge=True)
oat_query = self.inp.get_query_sets('oat',
group_by='hour',
group_by_aggregation=Avg,
exclude={'value': None},
wrap_for_merge=True)
self.out.log("Getting unit conversions.", logging.INFO)
base_topic = self.inp.get_topics()
meta_topics = self.inp.get_topics_meta()
load_unit = meta_topics['load'][base_topic['load'][0]]['unit']
self.out.log("Convert loads from [{}] to [kW].".format(load_unit),
logging.INFO)
load_convertfactor = cu.getFactor_powertoKW(load_unit)
temperature_unit = meta_topics['oat'][base_topic['oat'][0]]['unit']
self.out.log(
"Convert temperatures from [{}] to [F].".format(temperature_unit),
logging.INFO)
# Match the values by timestamp
merged_load_oat = self.inp.merge(load_query, oat_query)
load_values = []
oat_values = []
datetime_values = []
for x in merged_load_oat:
if temperature_unit == 'celcius':
convertedTemp = cu.convertCelciusToFahrenheit(x['oat'][0])
elif temperature_unit == 'kelvin':
convertedTemp = cu.convertKelvinToCelcius(
cu.convertCelciusToFahrenheit(x['oat'][0]))
else:
convertedTemp = x['oat'][0]
load_values.append(x['load'][0] *
load_convertfactor) #Converted to kWh
oat_values.append(convertedTemp)
datetime_values.append(x['time'])
indexList = {}
indexList['trainingStart'] = ttow.findDateIndex(
datetime_values, self.baseline_start)
self.out.log('@trainingStart ' + str(indexList['trainingStart']),
logging.INFO)
indexList['trainingStop'] = ttow.findDateIndex(datetime_values,
self.baseline_stop)
self.out.log('@trainingStop ' + str(indexList['trainingStop']),
logging.INFO)
indexList['predictStart'] = ttow.findDateIndex(datetime_values,
self.savings_start)
self.out.log('@predictStart ' + str(indexList['predictStart']),
logging.INFO)
indexList['predictStop'] = ttow.findDateIndex(datetime_values,
self.savings_stop)
self.out.log('@predictStop ' + str(indexList['predictStop']),
logging.INFO)
for indx in indexList.keys():
if indexList[indx] == None:
self.out.log("Date not found in the datelist", logging.WARNING)
# Break up data into training and prediction periods.
timesTrain = datetime_values[
indexList['trainingStart']:indexList['trainingStop']]
timesPredict = datetime_values[
indexList['predictStart']:indexList['predictStop']]
valsTrain = load_values[
indexList['trainingStart']:indexList['trainingStop']]
valsActual = load_values[
indexList['predictStart']:indexList['predictStop']]
oatsTrain = oat_values[
indexList['trainingStart']:indexList['trainingStop']]
oatsPredict = oat_values[
indexList['predictStart']:indexList['predictStop']]
# Generate other information needed for model.
timeStepMinutes = (timesTrain[1] - timesTrain[0]).total_seconds() / 60
# TODO: Should this be calculated in the utility function
binCt = 6 # TODO: Allow caller to pass this in as an argument.
# Form the temperature-time-of-week model.
self.out.log("Finding baseline model", logging.INFO)
ttowModel = ttow.formModel(timesTrain, oatsTrain, valsTrain,
timeStepMinutes, binCt)
# Apply the model.
self.out.log("Applying baseline model", logging.INFO)
valsPredict = ttow.applyModel(ttowModel, timesPredict, oatsPredict)
# Output for scatter plot
prevSum = 0
for ctr in range(len(timesPredict)):
# Calculate cumulative savings.
prevSum += (valsPredict[ctr] - valsActual[ctr])
local_time = str(
self.inp.localize_sensor_time(base_topic['load'][0],
timesPredict[ctr]))
self.out.insert_row(
"DayTimeTemperatureModel", {
"datetimeValues": local_time,
"measured": valsActual[ctr],
"predicted": valsPredict[ctr],
"cumulativeSum": prevSum
})