def test_load_and_trim_time_matrix_rows(sectors):
'''
Test trimmed time matrix rows are as expected
'''
matrix = io.load_travel_time()
trimmed = io.trim_matrix(matrix, sectors)
assert sectors == list(trimmed.index.values)
def test_load_and_trim_time_matrix_cols(sectors):
'''
Test trimmed time matrix columns are as expected
'''
matrix = io.load_travel_time()
trimmed = io.trim_matrix(matrix, sectors)
assert sectors == list(trimmed.columns)
def test_load_and_trim_time_matrix(sectors, expected_shape):
'''
Test trimmed time matrix shape is as expected
'''
matrix = io.load_travel_time()
trimmed = io.trim_matrix(matrix, sectors)
assert expected_shape == trimmed.shape
def test_capacity_one_cost(warehouse, selected_sectors, expected_cost):
'''
Test single occupancy cost (warehouse->city_i, city_i<-warehouse)
'''
matrix = io.trim_matrix(io.load_travel_distance(), selected_sectors)
cost = cn.single_capacity_cost(warehouse, matrix)
#allowing for minor floating point difference
assert pytest.approx(cost) == expected_cost
def single_replication(self):
'''
runs a single replication of the model.
sample covid positive patients that need transport
constructs routes to transport each patient to the hospital
for different capacities.
'''
#step 1: sample n patients that need transport
patients = self._sample_patients(self._params.n_patients)
#step 2: sample which patients need transport
transport_patients = self._sample_transport(self._params.p_transport,
self._params.n_patients)
#step 3: sample covid-19 positive patients
covid_patients = self._sample_covid(self._params.p_positive,
self._params.n_patients)
#step 4: covid_positive patients that need transport
covid_patient_mask = transport_patients * covid_patients
covid_patient_sectors = patients[covid_patient_mask.astype(bool)]
#step 5: unit demand (simplification at this point)
demand = demand_by_sector(self._params.warehouse,
covid_patient_sectors)
#step 6: trim the matrix to help with debug and efficiency
unique_sectors = list(demand.keys())
if self._params.warehouse not in unique_sectors:
unique_sectors.append(self._params.warehouse)
#6.1 create smaller matrix using unique sectors.
sub_matrix = io.trim_matrix(self._params.cost_matrix, unique_sectors)
#6.2 clone areas with multiple demands
#This because sector demand may exceed vehicle_capacity.
#This workaround allows multiple trips to same postcode sector
demand, sub_matrix = clone_sectors_with_multiple_demands(
demand, sub_matrix, intra_sector_travel_cost=5.0)
#6.3 sort columns and index (incase enforcing symmetry)
sub_matrix = sort_cost_matrix_lexographically(sub_matrix)
#step 7: for each capacity, create routes and record total cost.
costs = {}
costs['capacity_1'] = c.single_capacity_cost(self._params.warehouse,
sub_matrix)
for capacity in self._params.vehicle_capacities:
costs[f'capacity_{capacity}'] = []
journeys = self._solver.solve(vehicle_capacity=capacity,
matrix=sub_matrix,
demand=demand)
costs[f'capacity_{capacity}'] = self.total_routing_cost(
journeys, sub_matrix)
return costs
def test_travel_distance_cost(city_1, city_2, expected_cost):
'''
Test costs are symmetric.
'''
df = io.load_travel_distance()
sectors = ['L1', 'L10', 'L100','L101', 'L102']
trimmed = io.trim_matrix(df, sectors)
print(trimmed)
cost = io.travel_cost(city_1, city_2, trimmed)
assert expected_cost == cost