def test_sum(self):
males_under_5, males_under_5_moe = 10154024, 3778
females_under_5, females_under_5_moe = 9712936, 3911
self.assertEqual(
census_data_aggregator.approximate_sum(
(males_under_5, males_under_5_moe),
(females_under_5, females_under_5_moe)),
(19866960, 5437.757350231803))
# With multiple zeros
self.assertEqual(
census_data_aggregator.approximate_sum([0.0, 22], [0, 22], [0, 29],
[41, 37]),
(41, 47.01063709417264))
def _aggregate(group_df):
"""
The function that aggregates each group
"""
out = {}
for col in data_columns:
# The name of the error column (if it exists)
error_col = f"{col}_moe"
# remove any NaN rows
subset = group_df.dropna(subset=[col], how="any")
# aggregat if we had any rows left
if len(subset):
# column values, margin of error (if it exists)
args = np.column_stack([
subset[col],
subset.get(error_col, np.zeros(len(subset)))
])
# do the aggregation
aggval, moe = cda.approximate_sum(*args)
else:
aggval = moe = np.nan
# store
out[col] = aggval
if error_col in subset.columns:
out[f"{col}_moe"] = moe
out["geometry"] = group_df.geometry.unary_union
return pd.Series(out)
def calculate_total_bachelors_or_higher_moe(row):
if row['female_bachelors_or_higher_moe'] in list(MOE_MAP.values()):
value = sum([
row['female_bachelors_or_higher'],
row['male_bachelors_or_higher']
])
moe = None
value, moe = census_data_aggregator.approximate_sum(
(row['female_bachelors_degree'],
row['female_bachelors_degree_moe']),
(row['female_masters_degree'],
row['female_masters_degree_moe']),
(row['female_professional_school_degree'],
row['female_professional_school_degree_moe']),
(row['female_doctorate_degree'],
row['female_doctorate_degree_moe']),
(row['male_bachelors_degree'],
row['male_bachelors_degree_moe']),
(row['male_masters_degree'], row['male_masters_degree_moe']),
(row['male_professional_school_degree'],
row['male_professional_school_degree_moe']),
(row['male_doctorate_degree'],
row['male_doctorate_degree_moe']))
row['total_bachelors_or_higher'] = value
row['total_bachelors_or_higher_moe'] = moe
return row
def test_sum(self):
males_under_5, males_under_5_moe = 10154024, 3778
females_under_5, females_under_5_moe = 9712936, 3911
self.assertEqual(
census_data_aggregator.approximate_sum(
(males_under_5, males_under_5_moe),
(females_under_5, females_under_5_moe)),
(19866960, 5437.757350231803))
# With multiple zeros
self.assertEqual(
census_data_aggregator.approximate_sum([0.0, 22], [0, 22], [0, 29],
[41, 37]),
(41, 47.01063709417264))
# From the ACS handbook examples
single_women = ((135173, 3860), (43104, 2642), (24842, 1957))
self.assertEqual(census_data_aggregator.approximate_sum(*single_women),
(203119, 5070.4647715963865))
def approximate_sum(row, cols):
"""
Apply this function to a DataFrame, summing over the columns specified.
Note
----
This assumes a margin of error column exists
"""
args = [(row[col], row[f"{col}_moe"]) for col in cols]
return pd.Series(cda.approximate_sum(*args))
def test_sum_ch8(self):
# Never-married female characteristics from Table 8.1
nmf_fairfax = (135173, 3860)
nmf_arlington = (43104, 2642)
nmf_alexandria = (24842, 1957)
# Calculate aggregate pop and MOE
agg_pop, agg_moe = census_data_aggregator.approximate_sum(
nmf_fairfax, nmf_arlington, nmf_alexandria)
self.assertEqual(agg_pop, 203119)
self.assertAlmostEqual(agg_moe, 5070, places=0)
def calculate_asians_moe(row):
pprint(row)
if row['asian_alone_moe'] in list(MOE_MAP.values()):
value = sum([row['asian_alone'], row['native_hawaiian_and_pacific_islander']])
moe = None
value, moe = census_data_aggregator.approximate_sum(
(row['native_hawaiian_and_pacific_islander'], row['native_hawaiian_and_pacific_islander_moe']),
(row['asian_alone'], row['asian_alone_moe']),
)
row['asians_all'] = value
row['asians_all_moe'] = moe
return row
def calculate_moe(row):
pprint(row)
if row['internet_without_subscription_moe'] in list(MOE_MAP.values()):
value = sum([row['internet_without_subscription'], row['no_internet']])
moe = None
value, moe = census_data_aggregator.approximate_sum(
(row['internet_without_subscription'], row['internet_without_subscription_moe']),
(row['no_internet'], row['no_internet_moe']),
)
row['total_no_internet_and_no_subscription'] = value
row['total_no_internet_and_no_subscription_moe'] = moe
return row
def calculate_other_moe(row):
pprint(row)
if row['american_indian_and_alaska_native_moe'] in list(MOE_MAP.values()):
value = sum([row['american_indian_and_alaska_native'], row['other_alone'], row['two_or_more_races']])
moe = None
value, moe = census_data_aggregator.approximate_sum(
(row['american_indian_and_alaska_native'], row['american_indian_and_alaska_native_moe']),
(row['other_alone'], row['other_alone_moe']),
(row['two_or_more_races'], row['two_or_more_races_moe']),
)
row['other_all'] = value
row['other_all_moe'] = moe
return row
def test_proportion_ch8(self):
# Total females aged 15 and older from Table 8.4
tf15_fairfax = (466037, 391)
tf15_arlington = (97360, 572)
tf15_alexandria = (67101, 459)
# Aggregate the values and MOEs
denominator = census_data_aggregator.approximate_sum(
tf15_fairfax, tf15_arlington, tf15_alexandria)
numerator = (203119, 5070)
# Calculate the proportion and its MOE
proportion, moe = census_data_aggregator.approximate_proportion(
numerator, denominator)
self.assertAlmostEqual(proportion, 0.322, places=3)
self.assertAlmostEqual(moe, 0.008, places=3)
def test_ratio_ch8(self):
# Never-married Males from table 8.5
nmm_fairfax = (156720, 4222)
nmm_arlington = (44613, 2819)
nmm_alexandria = (25507, 2259)
# Aggregate the values and MOEs
numerator = census_data_aggregator.approximate_sum(
nmm_fairfax, nmm_arlington, nmm_alexandria)
denominator = (203119, 5070)
# Calculate the proportion and its MOE
ratio, moe = census_data_aggregator.approximate_ratio(
numerator, denominator)
self.assertAlmostEqual(ratio, 1.117, places=3)
self.assertAlmostEqual(moe, 0.039, places=3)
def calculate_moe(row):
# our custom groups
if row['us_citizen_by_naturalization_moe'] in list(
MOE_MAP.values()):
value = sum([
row['us_citizen_by_naturalization'],
row['us_citizen_born_abroad_american_parents'],
row['us_citizen_born_puertorico_or_us_island_alone'],
row['us_citizen_born_us']
])
moe = None
value, moe = census_data_aggregator.approximate_sum(
(row['us_citizen_by_naturalization'],
row['us_citizen_by_naturalization_moe']),
(row['us_citizen_born_abroad_american_parents'],
row['us_citizen_born_abroad_american_parents_moe']),
(row['us_citizen_born_puertorico_or_us_island_alone'],
row['us_citizen_born_puertorico_or_us_island_moe']),
(row['us_citizen_born_us'], row['us_citizen_born_us_moe']),
)
row['us_citizen_total'] = value
row['us_citizen_total_moe'] = moe
return row