def test_pearsonr_difference_significance(self):
# go through a few examples (calculations done using http://www.quantpsy.org/corrtest/corrtest.htm)
r_a = 0.3639
n_a = 91
r_b = 0.0205
n_b = 63
p = 2 * 0.01556
self.assertAlmostEqual(
p,
stats.pearsonr_difference_significance(r_a, n_a, r_b, n_b),
places=3,
)
r_a = 0.3
n_a = 200
r_b = 0.1
n_b = 100
p = 2 * 0.04585
self.assertAlmostEqual(
p,
stats.pearsonr_difference_significance(r_a, n_a, r_b, n_b),
places=3,
)
r_a = 0.7
n_a = 30
r_b = -0.3
n_b = 10
p = 2 * 0.00276
self.assertAlmostEqual(
p,
stats.pearsonr_difference_significance(r_a, n_a, r_b, n_b),
places=3,
)
r_a = -0.1
n_a = 20
r_b = 0.4
n_b = 4
p = 2 * 0.30529
self.assertAlmostEqual(
p,
stats.pearsonr_difference_significance(r_a, n_a, r_b, n_b),
places=3,
)
def test_pearsonr_difference_significance(self):
# go through a few examples (calculations done using http://www.quantpsy.org/corrtest/corrtest.htm)
r_a = 0.3639
n_a = 91
r_b = 0.0205
n_b = 63
p = 2*0.01556
self.assertAlmostEqual(
p, stats.pearsonr_difference_significance(r_a, n_a, r_b, n_b),
places=3,
)
r_a = 0.3
n_a = 200
r_b = 0.1
n_b = 100
p = 2*0.04585
self.assertAlmostEqual(
p, stats.pearsonr_difference_significance(r_a, n_a, r_b, n_b),
places=3,
)
r_a = 0.7
n_a = 30
r_b = -0.3
n_b = 10
p = 2*0.00276
self.assertAlmostEqual(
p, stats.pearsonr_difference_significance(r_a, n_a, r_b, n_b),
places=3,
)
r_a = -0.1
n_a = 20
r_b = 0.4
n_b = 4
p = 2*0.30529
self.assertAlmostEqual(
p, stats.pearsonr_difference_significance(r_a, n_a, r_b, n_b),
places=3,
)
axs[AXES[expt.id]].fill_between(
t, lbs, ubs, color=EARLY_LATE_COLORS[label], alpha=ALPHA
)
early_late_handles.append(handle)
early_late_labels.append(label)
if label == 'early':
# store these for later so we can compare them to the lates
early_correlations = correlations
early_ns = ns
elif label == 'late':
# calculate significance between early and late correlations
p_vals = []
for r_1, n_1, r_2, n_2 in zip(early_correlations, early_ns, correlations, ns):
p_vals.append(stats.pearsonr_difference_significance(r_1, n_1, r_2, n_2))
ax = axs_twin[AXES[expt.id] - 2]
ax.plot(t, p_vals, c='k', ls='-', lw=2)
ax.axhline(0.05, c='k', ls='--')
ax.set_ylim(0, 0.5)
axs_twin[AXES[expt.id] - 2].legend(early_late_handles, early_late_labels, loc='best')
axs[0].legend(wind_speed_handles, wind_speed_labels, loc='best')
axs[0].set_title('Concentration/heading\npartial correlations')
axs[1].set_ylim(0, 1)
axs[1].legend(wind_speed_handles, wind_speed_labels, loc='best')
axs[1].set_title('P-values')
alpha=ALPHA)
early_late_handles.append(handle)
early_late_labels.append(label)
if label == 'early':
# store these for later so we can compare them to the lates
early_correlations = correlations
early_ns = ns
elif label == 'late':
# calculate significance between early and late correlations
p_vals = []
for r_1, n_1, r_2, n_2 in zip(early_correlations, early_ns,
correlations, ns):
p_vals.append(
stats.pearsonr_difference_significance(
r_1, n_1, r_2, n_2))
ax = axs_twin[AXES[expt.id] - 2]
ax.plot(t, p_vals, c='k', ls='-', lw=2)
ax.axhline(0.05, c='k', ls='--')
ax.set_ylim(0, 0.5)
axs_twin[AXES[expt.id] - 2].legend(early_late_handles,
early_late_labels,
loc='best')
axs[0].legend(wind_speed_handles, wind_speed_labels, loc='best')
axs[0].set_title('Concentration/heading\npartial correlations')
axs[1].set_ylim(0, 1)
