def test_generate_household_sizes_from_fixed_pop_size(self):
"""
Test generate_household_sizes_from_fixed_pop_size the test data is
specifically crafted to execute all conditional branches of the method.
Returns:
None
"""
even_dist = {1: 0.2, 2: 0.2, 3: 0.2, 4: 0.2, 5: 0.2}
# 900 is divisble by the expected value (3.0) but 901 is not
# this creates test cases for N_gen = N and N_gen < N condition
for i in [900, 901]:
hh = sp.generate_household_sizes_from_fixed_pop_size(
N=i, hh_size_distr=even_dist)
# verify the total number of people matches N
self.assertEqual(i,
sum([(n + 1) * hh[n] for n in range(0, len(hh))]))
# verify distribution
self.verify_buckets(even_dist.values(), hh)
# slightly modify the distribution to create expected value = 2.91 which will round down to 2.9
# and create N_gen > N condition
uneven_dist = {1: 0.2, 2: 0.2, 3: 0.2, 4: 0.29, 5: 0.11}
hh2 = sp.generate_household_sizes_from_fixed_pop_size(
N=900, hh_size_distr=uneven_dist)
self.assertEqual(900,
sum([(n + 1) * hh2[n] for n in range(0, len(hh2))]))
self.verify_buckets(uneven_dist.values(), hh2)
def test_generate_all_households(location='seattle_metro',
state_location='Washington',
country_location='usa'):
N = 1000
household_size_distr = sp.get_household_size_distr(datadir, location,
state_location,
country_location)
hh_sizes = sp.generate_household_sizes_from_fixed_pop_size(
N, household_size_distr)
hha_brackets = sp.get_head_age_brackets(datadir,
country_location=country_location)
hha_by_size_counts = sp.get_head_age_by_size_distr(
datadir, country_location=country_location)
age_brackets_filepath = sp.get_census_age_brackets_path(
datadir, state_location, country_location)
age_brackets = sp.get_age_brackets_from_df(age_brackets_filepath)
age_by_brackets_dic = sp.get_age_by_brackets_dic(age_brackets)
contact_matrix_dic = sp.get_contact_matrix_dic(
datadir, sheet_name='United States of America')
single_year_age_distr = {}
for n in range(101):
single_year_age_distr[n] = float(1.0 / 101.0)
homes_dic, homes = sp.generate_all_households(
N, hh_sizes, hha_by_size_counts, hha_brackets, age_brackets,
age_by_brackets_dic, contact_matrix_dic, single_year_age_distr)
assert homes_dic, homes is not None
def test_generate_household_sizes_from_fixed_pop_size(location='seattle_metro', state_location='Washington',
country_location='usa'):
household_size_distr = sp.get_household_size_distr(datadir, location, state_location, country_location)
Nhomes = 1000
hh_sizes = sp.generate_household_sizes_from_fixed_pop_size(Nhomes, household_size_distr)
assert len(hh_sizes) == 7
def test_custom_household_size_distro_honored(self):
"""
This methods checks results from
generate_household_sizes_from_fixed_pop_size with customized
distribution. It checks that the most common household size should be
the size with the highest probability and also uses
verify_portion_honored method for validation logic.
Returns:
None
"""
self.is_debugging = False
custom_distro = {
1: 0.25,
2: 0.075,
3: 0.10,
4: 0.30,
5: 0.05,
6: 0.05,
7: 0.175
}
hh_sizes = sp.generate_household_sizes_from_fixed_pop_size(
500, custom_distro)
hh_size_list = list(hh_sizes) # Comes as np.ndarray
fewest_houses = min(hh_size_list)
fewest_index = hh_size_list.index(fewest_houses)
most_houses = max(hh_size_list)
most_index = hh_size_list.index(most_houses)
highest_probability = max(custom_distro.values())
lowest_probability = min(custom_distro.values())
most_houses_probability = custom_distro[most_index +
1] # hh_distro is 1 indexed
fewest_houses_probability = custom_distro[fewest_index + 1]
self.assertEqual(
highest_probability,
most_houses_probability,
msg=
"The most common household size should be the size with the highest probability"
)
prob_bucket_list = list(custom_distro.values())
self.verify_portion_honored(probability_buckets=prob_bucket_list,
count_buckets=hh_size_list,
portion=0.25)
self.verify_portion_honored(probability_buckets=prob_bucket_list,
count_buckets=hh_size_list,
portion=0.2)
self.verify_portion_honored(probability_buckets=prob_bucket_list,
count_buckets=hh_size_list,
portion=0.1)
print(household_size_distr)
Nhomes_to_sample_smooth = 100000
hh_sizes = sp.generate_household_sizes(Nhomes_to_sample_smooth,
household_size_distr)
totalpop = sp.get_totalpopsize_from_household_sizes(hh_sizes)
# hh_sizes = sp.generate_household_sizes(Nhomes,household_size_distr)
syn_ages, syn_sexes = sp.get_usa_age_sex_n(location, state_location,
totalpop)
syn_age_count = Counter(syn_ages)
syn_age_distr = sp.norm_dic(Counter(syn_ages))
N = Nhomes
hh_sizes = sp.generate_household_sizes_from_fixed_pop_size(
N, household_size_distr)
totalpop = sp.get_totalpopsize_from_household_sizes(hh_sizes)
print(totalpop, 'pop')
hha_df = sp.get_household_head_age_by_size_df(datadir, country_location,
use_bayesian)
hha_brackets = sp.get_head_age_brackets(datadir, country_location,
use_bayesian)
hha_by_size = sp.get_head_age_by_size_distr(datadir, country_location,
use_bayesian)
homes_dic, homes = sp.generate_all_households(hh_sizes, hha_by_size,
hha_brackets, age_brackets,
age_by_brackets_dic,
contact_matrix_dic,
