def get_old_and_new_dist_shape():
old_new_dist = pd.read_csv(get_data_dir('raw', 'district_relation.csv'))
old_new_dist_map = old_new_dist[[
'DIST2012', 'NEW_DIST_ID_2'
]].set_index('DIST2012').to_dict()['NEW_DIST_ID_2']
pop_dense = pd.read_csv(
get_data_dir('raw', 'district_pop_dens_friction.csv'))
old_new_dist = old_new_dist.merge(pop_dense, on='DIST2012')
# p = gpd.read_file('../data/geo2_zw2012.shp')
shape = p = gpd.read_file(
get_data_dir('raw', 'shapefiles', 'new_districts', 'ZWE_adm2.shp'))
shape['total_population'] = shape['ID_2'].map(
old_new_dist.groupby('NEW_DIST_ID_2')['total_pop'].sum())
return old_new_dist, shape
def set_intra_district_decreased_mobility_rates(self, intra_district_decreased_mobility_rates_file):
self.LOCKDOWN_DECREASED_MOBILITY_RATE = pd.read_csv(
get_data_dir('preprocessed', 'mobility', intra_district_decreased_mobility_rates_file),
index_col=0
)['pctdif_distance']
self.LOCKDOWN_DECREASED_MOBILITY_RATE = {
self.DISTRICT_NAME_TO_ID[i]: j for i, j in self.LOCKDOWN_DECREASED_MOBILITY_RATE.items()
}
def set_new_district_seed(self, seed_infected):
# NOTE: This should be updated when the admin level is changed.
# model.params.DISTRICT_NAME_TO_ID['d_1'] -> 0 Bulawayo
# model.params.DISTRICT_NAME_TO_ID['d_2'] -> 11 Harare
# model.params.DISTRICT_NAME_TO_ID['d_18'] -> 9 Goromonzi
with open(get_data_dir('preprocessed', 'line_list', 'latest_line_list.pickle'), 'rb') as fl:
infected_count = pickle.load(fl)
self.DISTRICT_ID_INFECTED_COUNT = {self.DISTRICT_NAME_TO_ID.get(i): j for i, j in infected_count.items()}
self.DISTRICT_ID_INFECTED_PROB = pd.Series(self.DISTRICT_ID_INFECTED_COUNT)
self.DISTRICT_ID_INFECTED_PROB = (self.DISTRICT_ID_INFECTED_PROB / self.DISTRICT_ID_INFECTED_PROB.sum()).to_dict()
self.SEED_INFECT_DISTRICT_IDS = np.array([i for i in self.DISTRICT_ID_INFECTED_COUNT])
self.SEED_INFECT_AGE_MIN = 20
self.SEED_INFECT_AGE_MAX = 60
self.SEED_INFECT_NUM = seed_infected # 3 -> 66 for a 5% sample
self.SIMULATION_START_DATE = datetime(2020, 9, 1, 8)
self.data_file_name = get_data_dir('preprocessed', 'census', f'zimbabwe_expanded_census_consolidated_{self.data_sample_size}pct.pickle')
def scenario_handwashing_risk(self):
self.SCENARIO = 'HANDWASHING_RISK'
self.risk_data = pd.read_csv(
get_data_dir('preprocessed', 'risk', 'hw_and_severe_disease_risk.csv'))
self.DISTRICT_HW_RISK = self.risk_data[['ID_2', 'mean_hw_risk_pop_weighted']].copy()
self.MEAN_HW_RISK = self.DISTRICT_HW_RISK['mean_hw_risk_pop_weighted'].mean()
self.MIN_HW_RISK = self.DISTRICT_HW_RISK['mean_hw_risk_pop_weighted'].min()
self.DISTRICT_HW_RISK['ID_2'] = self.DISTRICT_HW_RISK['ID_2'].map(lambda x: f'd_{x}')
self.DISTRICT_HW_RISK['effective_hw_risk'] = self.DISTRICT_HW_RISK['mean_hw_risk_pop_weighted'].map(self.get_effective_R0)
self.DISTRICT_HW_RISK = self.DISTRICT_HW_RISK.set_index('ID_2').to_dict()['effective_hw_risk']
self.DISTRICT_SEVERE_DISEASE_RISK = self.risk_data[['ID_2', 'severe_covid_risk']].copy()
self.DISTRICT_SEVERE_DISEASE_RISK['severe_covid_risk'] = (
self.DISTRICT_SEVERE_DISEASE_RISK['severe_covid_risk'] / self.MEAN_HOSPITALIZATION_RATE
)
self.DISTRICT_SEVERE_DISEASE_RISK = self.DISTRICT_SEVERE_DISEASE_RISK.set_index('ID_2').to_dict()['severe_covid_risk']
def run_scenario(scenarioClass,
scenario_name,
sim_fname,
R0,
sample_size,
seed_num,
start_date=None,
timestep=timedelta(hours=4),
scaled_mobility=False):
import pickle
import sys
from covid19_abm.params import ParamsConfig
from covid19_abm.dir_manager import get_data_dir
# scenario_name = sys.argv[0]
# sim_fname = sys.argv[1].zfill(2)
# R0 = 1.9
# sample_size = 10
# seed_num = 90 # 6 -> 66 for 10% data sample
sim_fname = sim_fname.zfill(2)
if scaled_mobility:
sim_fname = f'scaled_{scenario_name}_{sim_fname}_R{R0}_samp{sample_size}_seed{seed_num}'
stay_duration_file = 'weekday_mobility_duration_count_df-new-district-scaled.pickle'
transition_probability_file = 'daily_region_transition_probability-new-district-scaled.csv'
else:
sim_fname = f'{scenario_name}_{sim_fname}_R{R0}_samp{sample_size}_seed{seed_num}'
stay_duration_file = 'weekday_mobility_duration_count_df-new-district.pickle'
transition_probability_file = 'daily_region_transition_probability-new-district-pre-lockdown.csv'
now = datetime.now().isoformat()
model_log_file = get_data_dir('logs',
f'model_log_file_{sim_fname}.{now}.log')
individual_log_file = get_data_dir(
'logs', f'individual_log_file_{sim_fname}.{now}.log')
params = ParamsConfig(
district='new',
data_sample_size=sample_size,
R0=R0,
normal_interaction_matrix_file=get_data_dir(
'raw', 'final_close_interaction_matrix_normal.xlsx'),
lockdown_interaction_matrix_file=get_data_dir(
'raw', 'final_close_interaction_matrix_lockdown.xlsx'),
stay_duration_file=get_data_dir('preprocessed', 'mobility',
stay_duration_file),
transition_probability_file=get_data_dir('preprocessed', 'mobility',
transition_probability_file),
timestep=timestep)
params.set_new_district_seed(seed_infected=seed_num)
model = scenarioClass(params,
model_log_file=model_log_file,
individual_log_file=individual_log_file)
if start_date is not None:
params.SIMULATION_START_DATE = start_date
model.scheduler.real_time = params.SIMULATION_START_DATE
model.load_agents(params.data_file_name,
size=None,
infect_num=params.SEED_INFECT_NUM)
# end_date = model.scheduler.real_time + timedelta(days=30 * 24, hours=4)
end_date = datetime(2021, 6, 1)
while model.scheduler.real_time <= end_date:
model.step()
if ((model.epidemic_state >= model.STATE_INFECTED) &
(model.epidemic_state < model.STATE_RECOVERED)).sum() == 0:
break
model_dump_file = get_data_dir(
'logs', f'model_dump_file_{sim_fname}.{now}.pickle')
with open(model_dump_file, 'wb') as fl:
pickle.dump(model, fl)
def scenario_test_interaction_matrix_sensitivity(self):
self.SCENARIO = 'INTERACTION_MATRIX_SENSITIVITY'
self.set_interaction_parameters(
get_data_dir('raw', 'sensitivity_interaction_matrix.xlsx'))