def test_can_sample_range(self):
attributes = {
'example_p': ProbabilityDistribution({
0: 0.01,
1: 0.01,
2.006: 0.01,
3: 0.01,
4: 0.01,
5: 0.01,
6: 0.88,
7: 0.01,
8: 0.01,
9: 0.01,
10: 0.01,
11: 0.01,
12: 0.01,
})
}
dwelling = Dwelling(attributes, self.mock_connection)
dwelling.outputs = {
'example': {
'type': 'numrange',
'sampling': True,
'distribution': 'example_p'
}
}
self.sampling_module.process(dwelling)
# Should take 95% interval by default,
# should round to 2 decimals.
self.assertEqual(str(dwelling.attributes['example']), '[2.01, 10]')
def test_cavity_walls(self):
# with cavity wall
attributes = {
'bouwjaar': 1920,
'woningtype': 'vrijstaand'
}
p_multiplier = 1.98
dwelling = Dwelling(attributes, self.mock_connection)
cavity_wall_dist = self.insulation_module.process_insulation_type(dwelling, 'cavity wall')
eligible_dwellings_cavity_wall_n = 3273065
measure_n = 76784 + 114914 + 117197 + 96150 + 131324 + 132769 + 159507 + 159080 + 214035 + 281276
p_measure = p_multiplier * measure_n / eligible_dwellings_cavity_wall_n
self.assertEqual(cavity_wall_dist.p((1.25, 2.175)), p_measure)
# without cavity wall
attributes = {
'bouwjaar': 1919,
'woningtype': 'vrijstaand'
}
dwelling = Dwelling(attributes, self.mock_connection)
cavity_wall_dist = self.insulation_module.process_insulation_type(dwelling, 'cavity wall')
self.assertEqual(cavity_wall_dist.mean, 0)
# with recent (probably already insulated) cavity wall
attributes = {
'bouwjaar': 2020,
'woningtype': 'vrijstaand'
}
dwelling = Dwelling(attributes, self.mock_connection)
cavity_wall_dist = self.insulation_module.process_insulation_type(dwelling, 'cavity wall')
self.assertEqual(cavity_wall_dist.mean, 0)
def test_updating_dwelling_with_placeholder_processed_by_doesnt_override(
self):
self.placeholder_dwelling.processed_by = ['BaseModule']
# dwelling with same vbo_id as placeholder
attributes = {'vbo_id': self.vbo_id}
dwelling = Dwelling(attributes, self.connection)
dwelling.processed_by = ['OtherModule']
self.region.add_dwelling(dwelling)
self.assertEqual(dwelling.processed_by, ['OtherModule', 'BaseModule'])
def test_raises_when_adding_existing_dwelling(self):
attributes_1 = {'vbo_id': '0363010000000002'}
dwelling_1 = Dwelling(attributes_1, self.connection)
attributes_2 = attributes_1.copy()
dwelling_2 = Dwelling(attributes_2, self.connection)
self.region.dwellings = [dwelling_1]
# Attempt to add the same dwelling.
add_dwelling_partial = partial(self.region.add_dwelling, dwelling_2)
self.assertRaises(ValueError, add_dwelling_partial)
def test_can_sample_boolean(self):
attributes = {
'example_p': 1
}
dwelling = Dwelling(attributes, self.mock_connection)
dwelling.outputs = {
'example': {
'type': 'boolean',
'sampling': True,
'distribution': 'example_p'
}
}
self.sampling_module.process(dwelling)
self.assertTrue(dwelling.attributes['example'])
def test_facade_includes_cavity_wall_values(self):
# has cavity wall
attributes = {
'bouwjaar': 1920,
'woningtype': 'vrijstaand'
}
# FACADE
# Applicable WoON distribution:
# 2.11: 0.548,
# ...
p_multiplier = 1.98
# For dwellings built in or before 2000, all measure years
# from 2010 to 2019 are applicable.
p_facade_measure_base_before_2000 = 35838 / 6591218 + 73097 / 6669286 + 76480 / 6739330 + 60548 / 6804459 + 84501 / 6870704 + 74448 / 6943943 + 75802 / 7027060 + 85442 / 7109692 + 100978 / 7189902 + 125197 / 7261671
p_facade_measure = p_multiplier * p_facade_measure_base_before_2000
eligible_dwellings_cavity_wall_n = 3273065
cavity_measure_n = 76784 + 114914 + 117197 + 96150 + 131324 + 132769 + 159507 + 159080 + 214035 + 281276
p_cavity_measure = p_multiplier * cavity_measure_n / eligible_dwellings_cavity_wall_n
dwelling = Dwelling(attributes, self.mock_connection)
self.insulation_module.process(dwelling)
# R-value is 2.11 if it was 2.11 to begin with (p=00.548)
# and no measures were taken after that.
self.assertAlmostEqual(dwelling.attributes['insulation_facade_r_dist'].p(2.11), 0.548 * (1 - p_facade_measure) * (1-p_cavity_measure))
def test_uses_woon_data_and_measures_for_buildings_before_2006(self):
# no cavity wall
attributes = {
'bouwjaar': 1919,
'woningtype': 'vrijstaand'
}
# FACADE
# Applicable WoON distribution:
# 0.36: 0.018,
# 0.43: 0.469,
# ...
p_multiplier = 1.98
# For dwellings built in or before 2000, all measure years
# from 2010 to 2019 are applicable.
p_facade_measure_base_before_2000 = 35838 / 6591218 + 73097 / 6669286 + 76480 / 6739330 + 60548 / 6804459 + 84501 / 6870704 + 74448 / 6943943 + 75802 / 7027060 + 85442 / 7109692 + 100978 / 7189902 + 125197 / 7261671
p_facade_measure = p_multiplier * p_facade_measure_base_before_2000
dwelling = Dwelling(attributes, self.mock_connection)
self.insulation_module.process(dwelling)
# R-value is 0.36 if it was 0.36 to begin with (p=0.469)
# and no measures were taken after that (p = 1 - p_measure).
self.assertAlmostEqual(dwelling.attributes['insulation_facade_r_dist'].p(0.36), 0.018 * (1 - p_facade_measure), places=4)
p_roof_measure_base_before_2000 = 115398 / 6591218 + 157347 / 6669286 + 195420 / 6739330 + 124267 / 6804459 + 155099 / 6870704 + 148447 / 6943943 + 148100 / 7027060 + 164024 / 7109692 + 199784 / 7189902 + 246325 / 7261671
p_roof_measure = p_multiplier * p_roof_measure_base_before_2000
self.assertAlmostEqual(dwelling.attributes['insulation_roof_r_dist'].p(0.39), 0.025 * (1 - p_roof_measure), places=4)
def test_raises_when_adding_dwelling_outside_region(self):
# dwelling with vbo_id that is not inside that region
attributes = {'vbo_id': '0363010000000002'}
dwelling = Dwelling(attributes, self.connection)
add_dwelling_partial = partial(self.region.add_dwelling, dwelling)
self.assertRaises(ValueError, add_dwelling_partial)
def test_can_replace_placeholders(self):
# dwelling with same vbo_id as placeholder
attributes = self.placeholder_attributes.copy()
dwelling = Dwelling(attributes, self.connection)
self.region.add_dwelling(dwelling)
# placeholder has been replaced
self.assertEqual(self.region.dwellings, [dwelling])
def test_saves_calculated_values(self):
attributes = {
'bouwjaar': 1920,
'woningtype': 'vrijstaand'
}
dwelling1 = Dwelling(attributes, self.mock_connection)
# identical dwelling
dwelling2 = Dwelling(attributes, self.mock_connection)
self.insulation_module.process(dwelling1)
# We monkeypatch the method,
# to see if it is called.
# If it is, it will fail with a TypeError.
self.insulation_module.process_insulation_type = lambda x, y: None
try:
self.insulation_module.process(dwelling2)
except TypeError as e:
self.fail(f'Should not rise TypeError "{e}"')
def test_updates_dwelling_with_placeholder_processed_by_when_adding(self):
self.placeholder_dwelling.processed_by = ['BaseModule']
# dwelling with same vbo_id as placeholder
attributes = {'vbo_id': self.vbo_id}
dwelling = Dwelling(attributes, self.connection)
self.region.add_dwelling(dwelling)
self.assertEqual(dwelling.processed_by, ['BaseModule'])
def test_raises_when_adding_dwelling_with_conflicting_information(self):
self.placeholder_dwelling.attributes['foo'] = 'bar'
# dwelling with same vbo_id as placeholder
attributes = {'vbo_id': self.vbo_id, 'foo': 'spam'}
dwelling = Dwelling(attributes, self.connection)
add_dwelling_partial = partial(self.region.add_dwelling, dwelling)
self.assertRaises(ValueError, add_dwelling_partial)
def test_can_sample_distribution(self):
attributes = {
'example_p': ProbabilityDistribution({
1: 0.5,
2: 0.5
})
}
dwelling = Dwelling(attributes, self.mock_connection)
dwelling.outputs = {
'example': {
'type': 'double precision',
'sampling': True,
'distribution': 'example_p'
}
}
self.sampling_module.process(dwelling)
# Should take mean by default
self.assertAlmostEqual(dwelling.attributes['example'], 1.5)
def test_updates_dwelling_with_placeholder_dwelling_values_when_adding(
self):
self.placeholder_dwelling.attributes['foo'] = 'bar'
# dwelling with same vbo_id as placeholder
attributes = {'vbo_id': self.vbo_id}
dwelling = Dwelling(attributes, self.connection)
self.region.add_dwelling(dwelling)
self.assertEqual(dwelling.attributes['foo'], 'bar')
def test_retains_info_when_adding_dwelling_with_same_information(self):
self.placeholder_dwelling.attributes['foo'] = 'bar'
# dwelling with same vbo_id as placeholder
attributes = {'vbo_id': self.vbo_id, 'foo': 'bar'}
dwelling = Dwelling(attributes, self.connection)
self.region.add_dwelling(dwelling)
self.assertEqual(dwelling.attributes, attributes)
def test_does_not_raise_when_adding_dwelling_with_same_information(self):
self.placeholder_dwelling.attributes['foo'] = 'bar'
# dwelling with same vbo_id as placeholder
attributes = {'vbo_id': self.vbo_id, 'foo': 'bar'}
dwelling = Dwelling(attributes, self.connection)
try:
self.region.add_dwelling(dwelling)
except ValueError:
self.fail('Should not raise ValueError')
def test_uses_national_average_when_no_other_averages(self):
pc6 = PC6('1000AA', self.connection)
# No average this time
pc6.attributes['energy_label_epi_log_avg'] = None
buurt = Buurt('BU0000000', self.connection)
buurt.attributes['energy_label_epi_log_avg'] = None
attributes = {
'vbo_id': '0003010000000001',
'bouwjaar': 2020,
'woningtype': 'tussenwoning'
}
dwelling = Dwelling(attributes, self.connection)
dwelling.regions['pc6'] = pc6
dwelling.regions['buurt'] = buurt
self.energy_label_prediction_module.process(dwelling)
# Exact value doesn't matter, but that it is different
# from previous results is.
self.assertTrue(dwelling.attributes['energy_label_epi_mean'] > 1.12)
def test_uses_building_code_for_new_buildings(self):
attributes = {
'bouwjaar': 2020,
'woningtype': 'vrijstaand'
}
dwelling = Dwelling(attributes, self.mock_connection)
self.insulation_module.process(dwelling)
self.assertEqual(dwelling.attributes['insulation_facade_r_dist'].p(4.5), 1)
self.assertEqual(dwelling.attributes['insulation_roof_r_dist'].p(6), 1)
self.assertEqual(dwelling.attributes['insulation_floor_r_dist'].p(3.5), 1)
self.assertEqual(dwelling.attributes['insulation_window_r_dist'].p(1/1.65), 1)
def test_predicts_epi(self):
attributes = {
'vbo_id': '0003010000000001',
'bouwjaar': 2020,
'woningtype': 'tussenwoning'
}
dwelling = Dwelling(attributes, self.connection)
pc6 = PC6('1000AA', self.connection)
pc6.attributes['energy_label_epi_log_avg'] = 0.1823215568
dwelling.regions['pc6'] = pc6
self.energy_label_prediction_module.process(dwelling)
self.assertAlmostEqual(dwelling.attributes['energy_label_epi_mean'], 1.1123215698958466, places=3)
epi_interval = dwelling.attributes['energy_label_epi_95']
self.assertAlmostEqual(epi_interval.lower, 0.7336717114532558)
self.assertAlmostEqual(epi_interval.upper, 1.6873401901864706)
self.assertEqual(dwelling.attributes['energy_label_class_mean'], 'B')
# A bit of an indirect way to check that this is the EnergyLabelClassRange
# from D to A.
self.energy_label_class_range_mock.assert_called_with('D', 'A', bounds='[]')
def test_uses_average_of_buurt_when_no_labels_in_pc6(self):
pc6 = PC6('1000AA', self.connection)
# No average this time
pc6.attributes['energy_label_epi_log_avg'] = None
buurt = Buurt('BU0000000', self.connection)
buurt.attributes['energy_label_epi_log_avg'] = 0.1823215568
attributes = {
'vbo_id': '0003010000000001',
'bouwjaar': 2020,
'woningtype': 'tussenwoning'
}
dwelling = Dwelling(attributes, self.connection)
dwelling.regions['pc6'] = pc6
dwelling.regions['buurt'] = buurt
self.energy_label_prediction_module.process(dwelling)
self.energy_label_prediction_module.process(dwelling)
# Should be the same as previous calculation.
self.assertAlmostEqual(dwelling.attributes['energy_label_epi_mean'], 1.1123215698958466, places=3)
def test_uses_building_code_and_measures_for_buildings_after_2006_windows(self):
attributes = {
'bouwjaar': 2008,
'woningtype': 'vrijstaand'
}
dwelling = Dwelling(attributes, self.mock_connection)
self.insulation_module.process(dwelling)
# WINDOWS
# base value: not from the building code,
# but for double glazing: 0.333
p_window_measure_2018 = 1.98 * 312337 / 7189902
p_window_measure_2019 = 1.98 * 376869 / 7261671
p_window_measure = p_window_measure_2018 + p_window_measure_2019
expected_window_mean = (1 - p_window_measure) * 0.333 + p_window_measure * (0.5 + 0.625)/2
self.assertAlmostEqual(dwelling.attributes['insulation_window_r_dist'].mean, expected_window_mean)
self.assertAlmostEqual(dwelling.attributes['insulation_window_r_dist'].p(0.333), 1 - p_window_measure)
def test_uses_building_code_and_measures_for_buildings_after_2006_facade(self):
# From 2006 and onwards,
# we don't have the WoON data anymore,
# so we use the building code.
# Easy example with just two applicable
# measure years (2018 and 2019).
attributes = {
'bouwjaar': 2008,
'woningtype': 'vrijstaand'
}
dwelling = Dwelling(attributes, self.mock_connection)
self.insulation_module.process(dwelling)
# FACADE
# Applicable building code:
# 2003: 2.5
# Measures R-value:
# 2018:
# 2.9: 11.4 / (11.4 + 14.5)
# 3.4: 14.5 / (11.4 + 14.5)
# 2019:
# 3.3: 10.2 / (10.2 + 11.8)
# 3.8: 11.8 / (10.2 + 11.8)
# Measures prob:
# multiplier: 1.98
# 2018 measures: 100978
# 2019 measures: 125197
# dwellings until 2008: 7189902
# dwellings until 2009: 7261671
p_facade_measure_2018 = 1.98 * 100978 / 7189902 # ~ 2.8%
p_facade_measure_2019 = 1.98 * 125197 / 7261671 # ~ 3.4%
p_facade_measure = p_facade_measure_2018 + p_facade_measure_2019
p_r_29 = 11.4 / (11.4 + 14.5)
p_r_34 = 14.5 / (11.4 + 14.5)
p_r_33 = 10.2 / (10.2 + 11.8)
p_r_38 = 11.8 / (10.2 + 11.8)
expected_facade_mean = 2.5 + p_facade_measure_2018 * (2.9 * p_r_29 + 3.4 * p_r_34) + p_facade_measure_2019 * (3.3 * p_r_33 + 3.8 * p_r_38)
self.assertAlmostEqual(dwelling.attributes['insulation_facade_r_dist'].mean, expected_facade_mean)
self.assertAlmostEqual(dwelling.attributes['insulation_facade_r_dist'].p(2.5), 1 - p_facade_measure)
def test_uses_building_code_and_measures_for_buildings_after_2006_roof(self):
attributes = {
'bouwjaar': 2008,
'woningtype': 'vrijstaand'
}
dwelling = Dwelling(attributes, self.mock_connection)
self.insulation_module.process(dwelling)
p_r_29 = 11.4 / (11.4 + 14.5)
p_r_34 = 14.5 / (11.4 + 14.5)
p_r_33 = 10.2 / (10.2 + 11.8)
p_r_38 = 11.8 / (10.2 + 11.8)
p_roof_measure_2018 = 1.98 * 199784 / 7189902
p_roof_measure_2019 = 1.98 * 246325 / 7261671
p_roof_measure = p_roof_measure_2018 + p_roof_measure_2019
expected_roof_mean = 2.5 + p_roof_measure_2018 * (2.9 * p_r_29 + 3.4 * p_r_34) + p_roof_measure_2019 * (3.3 * p_r_33 + 3.8 * p_r_38)
self.assertAlmostEqual(dwelling.attributes['insulation_roof_r_dist'].mean, expected_roof_mean)
self.assertAlmostEqual(dwelling.attributes['insulation_roof_r_dist'].p(2.5), 1 - p_roof_measure)
def pipeline(query, connection, fresh=False, N=None):
# set N = None to process full BAG.
# set fresh = True to delete previous results.
start_time = time.time()
print(f'fresh: {fresh} (if True, previous results will be deleted)')
# Also deletes existing `results' table
print("\nCreating table 'results'...")
create_results_table(fresh)
print("Adding primary key on vbo_id...")
make_primary_key('results', 'vbo_id')
print("\nInitiating modules...")
regional_modules = get_regional_modules(connection)
modules = get_modules(connection, regional_modules)
print("\nGetting dwellings...")
# We create a named server-side cursor:
# https://www.psycopg.org/docs/usage.html#server-side-cursors
# This keeps the memory usage down
# since it will only fetch about 2000 (see cursor.itersize)
# rows at a time into the Python memory.
# You don't need to close() this cursor afterwards (in fact
# the cursor disappears after a commit).
cursor = connection.cursor(name='pipeline-cursor')
cursor.execute(query)
# import pdb
# pdb.set_trace()
bag_count = 7892928
results_count_estimate = get_rowcount_estimate('results', connection)
print(f'Batch statistics:')
print(f' BAG entries: {bag_count}')
print(
f' estimate of current number of results (might be outdated): {results_count_estimate} ({results_count_estimate/bag_count*100:.2f}%)'
)
print(f' this batch: {"no number specified" if N is None else N}')
print('\nStarting processing...')
i = 0
for (vbo_id, pc6, oppervlakte, bouwjaar, woningtype, buurt_id) in cursor:
attributes = {
'vbo_id': vbo_id,
'pc6': pc6,
'oppervlakte': oppervlakte,
'bouwjaar': bouwjaar,
'woningtype': woningtype,
'buurt_id': buurt_id
}
dwelling = Dwelling(attributes, connection)
for module in modules:
module.process(dwelling)
dwelling.save()
i += 1
if i % 100 == 0:
print(f' processed dwelling: {i}', end='\r')
if i == N:
break
print("\n\nCommiting and closing...")
connection.commit()
connection.close()
print(
f'Processed {i:,} records in {(time.time() - start_time):.2f} seconds.'
)
def test_gets_both_energy_label_class_and_epi_imputed(self):
dwelling = Dwelling({'vbo_id': '0003010000000001'}, self.connection)
self.energy_label_module.process(dwelling)
self.assertEqual(dwelling.attributes['energy_label_class'], 'A')
self.assertEqual(dwelling.attributes['energy_label_epi'], 0.7)
def test_sets_to_none_when_no_label(self):
dwelling = Dwelling({'vbo_id': '0003010000000002'}, self.connection)
self.energy_label_module.process(dwelling)
self.assertEqual(dwelling.attributes['energy_label_class'], None)
self.assertEqual(dwelling.attributes['energy_label_epi'], None)