def test__init__(self):
with self.assertRaises(ValueError):
bp.BranchProModel(0, [0])
with self.assertRaises(TypeError):
bp.BranchProModel('0', [1])
with self.assertRaises(ValueError):
bp.BranchProModel(0, 1)
def test_simulate(self):
branch_model_1 = bp.BranchProModel(2, np.array([1, 2, 3, 2, 1]))
simulated_sample_model_1 = branch_model_1.simulate(1, np.array([2, 4]))
new_simulated_sample_model_1 = branch_model_1.simulate(1, [0, 2, 4])
self.assertEqual(simulated_sample_model_1.shape, (2,))
self.assertEqual(new_simulated_sample_model_1.shape, (3,))
branch_model_2 = bp.BranchProModel(2, [1, 2, 3, 2, 1])
simulated_sample_model_2 = branch_model_2.simulate(1, [2, 4, 7])
self.assertEqual(simulated_sample_model_2.shape, (3,))
br_model3 = bp.BranchProModel(0, [1, 2])
br_model3.set_r_profile([3, 1], [1, 2], 3)
simulated_sample_model_3 = br_model3.simulate(1, [2, 4, 7])
self.assertEqual(simulated_sample_model_3.shape, (3,))
def test_set_serial_intervals(self):
br_model = bp.BranchProModel(0, [1, 2])
br_model.set_serial_intervals([1, 3, 2])
npt.assert_array_equal(
br_model.get_serial_intervals(),
np.array([1, 3, 2])
)
with self.assertRaises(ValueError):
br_model.set_serial_intervals((1))
def test_set_r_profile(self):
br_model1 = bp.BranchProModel(0, [1, 2])
br_model1.set_r_profile([1], [2])
npt.assert_array_equal(br_model1.get_r_profile(), np.array([0, 1]))
br_model2 = bp.BranchProModel(0, [1, 2])
br_model2.set_r_profile([3, 1], [1, 2], 3)
npt.assert_array_equal(br_model2.get_r_profile(), np.array([3, 1, 1]))
with self.assertRaises(ValueError):
br_model1.set_r_profile(1, [1])
with self.assertRaises(ValueError):
br_model1.set_r_profile([1], 1)
with self.assertRaises(ValueError):
br_model1.set_r_profile([0.5, 1], [1])
with self.assertRaises(ValueError):
br_model1.set_r_profile([1], [-1])
with self.assertRaises(ValueError):
br_model1.set_r_profile([1, 2], [2, 1])
def test_run(self):
br_pro_model = bp.BranchProModel(2, np.array([1, 2, 3, 2, 1]))
simulationController = bp.SimulationController(br_pro_model, 2, 7)
one_run_of_simulator = simulationController.run(1)
self.assertEqual(one_run_of_simulator.shape, (6, ))
def test_get_regime(self):
br_pro_model = bp.BranchProModel(2, np.array([1, 2, 3, 2, 1]))
simulationController = bp.SimulationController(br_pro_model, 2, 7)
regime = simulationController.get_regime()
npt.assert_array_equal(regime, np.arange(2, 8).astype(int))
def test_get_time_bounds(self):
br_pro_model = bp.BranchProModel(2, np.array([1, 2, 3, 2, 1]))
simulationController = bp.SimulationController(br_pro_model, 2, 7)
bounds = simulationController.get_time_bounds()
self.assertEqual(bounds, (2, 7))
def test_get_r_profile(self):
br_model1 = bp.BranchProModel(0, [1, 2])
br_model1.set_r_profile([1], [2])
npt.assert_array_equal(br_model1.get_r_profile(), np.array([0, 1]))
def test_get_serial_intervals(self):
br_model = bp.BranchProModel(0, [1, 2])
npt.assert_array_equal(
br_model.get_serial_intervals(), np.array([1, 2]))
# Generate synthetic data
num_timepoints = 30 # number of days for incidence data
# Build the serial interval w_s
ws_mean = 2.6
ws_var = 1.5**2
theta = ws_var / ws_mean
k = ws_mean / theta
w_dist = scipy.stats.gamma(k, scale=theta)
disc_w = w_dist.pdf(np.arange(1, num_timepoints+1))
# Simulate incidence data
initial_r = 3
serial_interval = disc_w
model = bp.BranchProModel(initial_r, serial_interval)
new_rs = [3, 0.5] # sequence of R_0 numbers
start_times = [0, 15] # days at which each R_0 period begins
model.set_r_profile(new_rs, start_times)
parameters = 10 # initial number of cases
times = np.arange(num_timepoints)
cases = model.simulate(parameters, times)
example_data = pd.DataFrame({
'Days': times,
'Incidence Number': cases,
'R_t': [np.nan] + list(model.get_r_profile())
})
sliders = ['epsilon', 'mean', 'stdev', 'tau', 'central_prob']
def update_simulation(self, new_init_cond, new_r0, new_r1, new_t1,
new_epsilon):
"""Run a simulation of the branchpro model at the given slider values.
Parameters
----------
new_init_cond
(int) updated position on the slider for the number of initial
cases for the Branch Pro model in the simulator.
new_r0
(float) updated position on the slider for the initial reproduction
number for the Branch Pro model in the simulator.
new_r1
(float) updated position on the slider for the second reproduction
number for the Branch Pro model in the simulator.
new_t1
(float) updated position on the slider for the time change in
reproduction numbers for the Branch Pro model in the simulator.
new_epsilon
(float) updated position on the slider for the constant of
proportionality between local and imported cases for the Branch Pro
model in the posterior.
Returns
-------
pandas.DataFrame
Simulations storage dataframe
"""
data = self.session_data.get('data_storage')
serial_interval = self.session_data.get(
'interval_storage').iloc[:, 0].values
if data is None:
raise dash.exceptions.PreventUpdate()
time_label, inc_label = (data.columns[0], 'Incidence Number')
times = data[time_label]
# Make a new dataframe to save the simulation result
simulations = data[[time_label]]
# Add the correct R profile to the branchpro model
if 'Imported Cases' in data.columns:
br_pro_model = bp.LocImpBranchProModel(new_r0, serial_interval,
new_epsilon)
br_pro_model.set_imported_cases(
times, data.loc[:, ['Imported Cases']].squeeze().tolist())
else:
br_pro_model = bp.BranchProModel(new_r0, serial_interval)
br_pro_model.set_r_profile([new_r0, new_r1], [0, new_t1])
# Generate one simulation trajectory from this model
simulation_controller = bp.SimulationController(
br_pro_model, min(times), max(times))
try:
sim_data = simulation_controller.run(new_init_cond)
except ValueError:
sim_data = -np.ones(max(times))
# Add data to simulations storage
sim_times = simulation_controller.get_regime()
simulations = pd.DataFrame({
time_label: sim_times,
inc_label: sim_data
})
return simulations