def plot(covariates,
cases,
fit_params,
filename,
definitions,
discharges=None):
'''Plot the given cases and the '''
dist_params = {
'self_excitation_shape': 2.6,
'self_excitation_scale': 2.5,
'discharge_excitation_shape': 2.6,
'discharge_excitation_scale': 2.5
}
intensity = likelihood.carehome_intensity(cases=cases,
fit_params=fit_params,
covariates=covariates,
dist_params=dist_params,
discharges=discharges)
fig, ax = plt.subplots()
ax.set_xlabel('Days')
ax.set_ylabel('Intensity')
ax_right = ax.twinx()
ax_right.set_ylabel('Number of events')
markers = 'ox+*^'
linestyles = ['-', '--', ':', '-.', (0, (4, 1, 1, 1, 1, 1))]
for column, (marker, linestyle, (label, original_cases)) in enumerate(
zip(markers, linestyles, definitions.items())):
colour = f'C{column}'
ax.plot(intensity[:, column], color=colour, ls=linestyle)
if original_cases:
ax_right.scatter(*zip(*original_cases),
color=colour,
marker=marker,
ls='None')
else:
marker = 'None'
ax.plot(nan, color=colour, marker=marker, label=label, ls=linestyle)
ax.set_ylim((0, None))
ax_right.set_ylim((0, None))
max_num_events = int(ax_right.get_ylim()[1])
num_num_event_ticks = 1 + min(5, max_num_events)
num_event_tick_spacing = max_num_events // (num_num_event_ticks - 1)
num_event_ticks = arange(0, max_num_events + 1, num_event_tick_spacing)
ax_right.set_yticks(num_event_ticks)
if len(definitions) > 4:
legend_columns = 2
else:
legend_columns = 1
ax.legend(loc='best', frameon=False, ncol=legend_columns)
fig.tight_layout()
fig.savefig(filename)
plt.close(fig)
def test_likelihood_performance(large_test_data, benchmark):
'''
Test the performance of the calculation of likelihood from the intensity
and case distribution.'''
_, cases, covariates, discharges = large_test_data
intensity = likelihood.carehome_intensity(fit_params=LARGE_FIT_PARAMS,
covariates=covariates,
cases=cases,
discharges=discharges,
dist_params=FULL_DIST_PARAMS)
benchmark(likelihood.likelihood, intensity, cases)
def test_carehome_intensity_no_discharges(small_cases, small_covariates):
'''Test that the behaviour of carehome_intensity in the case where
discharges are not considered.'''
_, cases = small_cases
_, covariates = small_covariates
fit_params_no_rh = {**SMALL_FIT_PARAMS, 'r_h': None}
intensity = likelihood.carehome_intensity(covariates=covariates,
cases=cases,
fit_params=fit_params_no_rh,
dist_params=SIMPLE_DIST_PARAMS)
assert_almost_equal(
intensity, [[1, 2, 2], [1.736, 2.184, 3.104], [2.277, 2.135, 3.364]],
decimal=3)
def test_intensity_performance_self(large_test_data, benchmark, use_cache):
'''
Test the performance of the intensity function with self-excitation
'''
_, cases, covariates, _ = large_test_data
if not use_cache:
# Writeable arrays are not cached
cases.flags.writeable = True
covariates.flags.writeable = True
kwargs = {
'fit_params': {
**LARGE_FIT_PARAMS, 'r_h': None
},
'covariates': covariates,
'cases': cases,
'dist_params': FULL_DIST_PARAMS
}
# Ensure that numba can jit the function before timing it
likelihood.carehome_intensity(**kwargs)
benchmark(likelihood.carehome_intensity, **kwargs)
def test_carehome_intensity_with_discharges(small_cases, small_covariates):
'''Test that the behaviour of carehome_intensity is correct in the case
where discharges are considered.'''
_, cases = small_cases
_, covariates = small_covariates
discharges = cases[::-1]
intensity = likelihood.carehome_intensity(covariates=covariates,
cases=cases,
fit_params=SMALL_FIT_PARAMS,
dist_params=SIMPLE_DIST_PARAMS,
discharges=discharges)
assert_almost_equal(
intensity, [[1, 2, 2], [2.077, 5.937, 3.217], [3.332, 11.240, 3.810]],
decimal=3)
def perform(self, node, inputs, outputs):
"""
Perform the Op; get the log-likelihood of the data given the inputs.
"""
start_index = 0
if self.fixed_r_c is None:
r_c = inputs[0][0]
start_index += 1
else:
r_c = self.fixed_r_c
if self.fixed_r_h is None:
r_h = inputs[0][start_index]
start_index += 1
else:
r_h = self.fixed_r_h
fit_params = {
'baseline_intensities': np.asarray(
inputs[0][start_index:]
),
'r_c': r_c,
'r_h': r_h
}
if (r_c == 0) and (r_h == 0):
intensity = likelihood.carehome_intensity_null(
covariates=self.covariates,
cases=self.cases,
fit_params=fit_params
)
else:
intensity = likelihood.carehome_intensity(
covariates=self.covariates,
cases=self.cases,
discharges=self.discharges,
fit_params=fit_params,
dist_params=self.dist_params
)
logl = likelihood.likelihood(intensity, self.cases)
if outputs is not None:
outputs[0][0] = np.array(logl)
else:
return logl