ht_ln_gt45 = mean_sd_plot(
axs[cur_row, 0], x, ht_mean_gt45, ht_sd_gt45,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[0], markevery=20))
st_ln_gt45 = mean_sd_plot(
axs[cur_row, 1], x, st_mean_gt45, st_sd_gt45,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[1], markevery=20))
gh_ln_gt45 = mean_sd_plot(
axs[cur_row, 2], x, gh_mean_gt45, gh_sd_gt45,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[2], markevery=20))
# plot SPM
ht_x_sig = sig_filter(ht_lt35_vs_gt35, x)
st_x_sig = sig_filter(st_lt35_vs_gt35, x)
gh_x_sig = sig_filter(gh_lt35_vs_gt35, x)
axs[cur_row, 0].plot(ht_x_sig,
np.repeat(ax_limits[cur_row][1], ht_x_sig.size),
color='k',
lw=2)
axs[cur_row, 1].plot(st_x_sig,
np.repeat(ax_limits[cur_row][1], st_x_sig.size),
color='k',
lw=2)
axs[cur_row, 2].plot(gh_x_sig,
np.repeat(ax_limits[cur_row][1], gh_x_sig.size),
color='k',
lw=2)
ht_ln_m = mean_sd_plot(
axs[cur_row, 0], x, ht_mean_m, ht_sd_m,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[0], markevery=20))
st_ln_m = mean_sd_plot(
axs[cur_row, 1], x, st_mean_m, st_sd_m,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[1], markevery=20))
gh_ln_m = mean_sd_plot(
axs[cur_row, 2], x, gh_mean_m, gh_sd_m,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[2], markevery=20))
# plot SPM
ht_x_sig = sig_filter(ht_f_vs_m, x)
st_x_sig = sig_filter(st_f_vs_m, x)
gh_x_sig = sig_filter(gh_f_vs_m, x)
axs[cur_row, 0].plot(ht_x_sig,
np.repeat(ax_limits[cur_row][1], ht_x_sig.size),
color='k',
lw=2)
axs[cur_row, 1].plot(st_x_sig,
np.repeat(ax_limits[cur_row][1], st_x_sig.size),
color='k',
lw=2)
axs[cur_row, 2].plot(gh_x_sig,
np.repeat(ax_limits[cur_row][1], gh_x_sig.size),
color='k',
lw=2)
color=color_map.colors[0],
marker=markers[0],
capsize=3)
axs_elev[cur_row].errorbar(max_pos,
st_gt45_max_mean,
yerr=st_gt45_max_sd,
color=color_map.colors[1],
marker=markers[1],
capsize=3)
# spm
lt35_vs_gt45 = spm_test(traj_st_lt35,
traj_st_gt45).inference(alpha,
two_tailed=True,
**infer_params)
x_sig = sig_filter(lt35_vs_gt45, x_elev)
axs_elev[cur_row].plot(x_sig,
np.repeat(spm_y[activity.lower()], x_sig.size),
color='k')
print(activity)
print(extract_sig(lt35_vs_gt45, x_elev))
if idx == 0:
leg_elev_mean.append(st_lt35_ln[0])
leg_elev_mean.append(st_gt45_ln[0])
for idx_act, (activity, activity_df) in enumerate(
db_er_endpts.groupby('Activity', observed=True)):
lt35_df = activity_df[activity_df['age_group'] == '<35']
gt45_df = activity_df[activity_df['age_group'] == '>45']
two_tailed=True,
**infer_params)
# plot mean +- sd
cur_row = act_row[activity.lower()]
old_shr_ln = mean_sd_plot(
axs[cur_row], x, old_shr_mean, old_shr_sd,
dict(color=color_map.colors[0], alpha=0.25),
dict(color=color_map.colors[0], marker=markers[0], markevery=20))
new_shr_ln = mean_sd_plot(
axs[cur_row], x, new_shr_mean, new_shr_sd,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[0], markevery=20))
# plot spm
x_sig = sig_filter(spm_res, x)
axs[cur_row].plot(x_sig, np.repeat(12, x_sig.size), color='k', lw=2)
if idx == 0:
leg_mean.append(old_shr_ln[0])
leg_mean.append(new_shr_ln[0])
# figure title and legend
plt.tight_layout(pad=0.25, h_pad=1.5, w_pad=0.5)
fig.suptitle('SHR Comparison', x=0.47, y=0.99, fontweight='bold')
plt.subplots_adjust(top=0.93)
leg_left = fig.legend(leg_mean[:2], ['Euler SHR', 'Contrib SHR'],
loc='upper left',
bbox_to_anchor=(0, 1),
ncol=1,
handlelength=1.5,
ht_lt35_vs_gt45 = spm_test(traj_ht_lt35[:, 1:],
traj_ht_gt45[:,
1:]).inference(alpha,
two_tailed=True,
**infer_params)
st_lt35_vs_gt45 = spm_test(traj_st_lt35[:, 1:],
traj_st_gt45[:,
1:]).inference(alpha,
two_tailed=True,
**infer_params)
gh_lt35_vs_gt45 = spm_test(traj_gh_lt35[:, 1:],
traj_gh_gt45[:,
1:]).inference(alpha,
two_tailed=True,
**infer_params)
ht_x_sig = sig_filter(ht_lt35_vs_gt45, x[1:])
st_x_sig = sig_filter(st_lt35_vs_gt45, x[1:])
gh_x_sig = sig_filter(gh_lt35_vs_gt45, x[1:])
axs[idx_act, 0].plot(ht_x_sig,
np.repeat(5, ht_x_sig.size),
color='k',
lw=2)
axs[idx_act, 1].plot(st_x_sig,
np.repeat(5, st_x_sig.size),
color='k',
lw=2)
axs[idx_act, 2].plot(gh_x_sig,
np.repeat(5, gh_x_sig.size),
color='k',
lw=2)
dict(color=color_map.colors[2], alpha=0.25),
dict(color=color_map.colors[2], marker=markers[2], markevery=20))
axs_elev[cur_row].errorbar(max_pos, gh_max_mean, yerr=gh_max_sd, color=color_map.colors[0],
marker=markers[0], capsize=3)
axs_elev[cur_row].errorbar(max_pos + 3, st_max_mean, yerr=st_max_sd, color=color_map.colors[1],
marker=markers[1], capsize=3)
axs_elev[cur_row].errorbar(max_pos - 3, ht_max_mean, yerr=ht_max_sd, color=color_map.colors[2],
marker=markers[2], capsize=3)
# spm
if activity.lower() == 'fe':
st_vs_gh = spm_test(traj_st - traj_gh, 0).inference(alpha, two_tailed=True, **infer_params)
else:
st_vs_gh = spm_test(-(traj_st - traj_gh), 0).inference(alpha, two_tailed=True, **infer_params)
x_sig = sig_filter(st_vs_gh, x_elev)
axs_elev[cur_row].plot(x_sig, np.repeat(spm_y[activity.lower()], x_sig.size), color='k')
print('Activity: {}'.format(activity))
print(extract_sig(st_vs_gh, x_elev))
print('P-values: ')
print(output_spm_p(st_vs_gh))
print('HT Max: {:.2f}'.format(np.abs(ht_max_mean)))
print('ST Max: {:.2f}'.format(np.abs(st_max_mean)))
print('GH Max: {:.2f}'.format(np.abs(gh_max_mean)))
print('Percentage: {:.2f}'.format(st_max_mean / ht_max_mean * 100))
if idx == 0:
leg_elev_mean.append(ht_ln[0])
leg_elev_mean.append(st_ln[0])
leg_elev_mean.append(gh_ln[0])
dict(color=color_map.colors[1], marker=markers[1], markevery=20))
st_contrib_ln_gt45 = mean_sd_plot(
axs_st[cur_row, 1], x, st_contrib_mean_gt45, st_contrib_sd_gt45,
dict(color=color_map.colors[3], alpha=0.25),
dict(color=color_map.colors[3], marker=markers[3], markevery=20))
gh_euler_ln_gt45 = mean_sd_plot(
axs_gh[cur_row, 0], x, gh_euler_mean_gt45, gh_euler_sd_gt45,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[1], markevery=20))
gh_contrib_ln_gt45 = mean_sd_plot(
axs_gh[cur_row, 1], x, gh_contrib_mean_gt45, gh_contrib_sd_gt45,
dict(color=color_map.colors[3], alpha=0.25),
dict(color=color_map.colors[3], marker=markers[3], markevery=20))
# plot spm
st_euler_x_sig = sig_filter(st_euler_spm, x)
st_contrib_x_sig = sig_filter(st_contrib_spm, x)
gh_euler_x_sig = sig_filter(gh_euler_spm, x)
gh_contrib_x_sig = sig_filter(gh_contrib_spm, x)
axs_st[cur_row, 0].plot(st_euler_x_sig,
np.repeat(st_lims[cur_row][1],
st_euler_x_sig.size),
color='k',
lw=2)
axs_st[cur_row, 1].plot(st_contrib_x_sig,
np.repeat(st_lims[cur_row][1],
st_euler_x_sig.size),
color='k',
lw=2)
axs_gh[cur_row, 0].plot(gh_euler_x_sig,
np.repeat(gh_lims[cur_row][1],
axs[cur_row, 1].errorbar(max_pos - 3,
gh_euler_max_mean,
yerr=gh_euler_max_sd,
color=color_map.colors[0],
marker=markers[0],
capsize=3)
axs[cur_row, 1].errorbar(max_pos,
gh_contrib_max_mean,
yerr=gh_contrib_max_sd,
color=color_map.colors[1],
marker=markers[1],
capsize=3)
# plot spm
st_x_sig = sig_filter(st_spm, x)
gh_x_sig = sig_filter(gh_spm, x)
axs[cur_row, 0].plot(st_x_sig,
np.repeat(2, st_x_sig.size),
color='k',
lw=2)
axs[cur_row, 1].plot(gh_x_sig,
np.repeat(0, gh_x_sig.size),
color='k',
lw=2)
print(activity)
print('Mean Diff at Max ST: {:.2f}'.format(st_contrib_max_mean -
st_euler_max_mean))
print('Mean Diff at Max GH: {:.2f}'.format(gh_contrib_max_mean -
gh_euler_max_mean))
print('ST HT Elevation angles')
# plot at max
axs[cur_row].errorbar(max_pos,
ht_euler_add_max_mean,
yerr=ht_euler_add_max_sd,
color=color_map.colors[0],
marker=markers[0],
capsize=3)
axs[cur_row].errorbar(max_pos - 3,
ht_contrib_max_mean,
yerr=ht_contrib_max_sd,
color=color_map.colors[1],
marker=markers[1],
capsize=3)
# plot spm
x_sig_euler_add = sig_filter(ht_euler_diff_vs_euler_add, x)
# x_sig_contribs = sig_filter(ht_euler_diff_vs_contribs, x)
axs[cur_row].plot(x_sig_euler_add,
np.repeat(spm_y[cur_row], x_sig_euler_add.size),
color=color_map.colors[0],
lw=2)
# axs[cur_row].plot(x_sig_contribs, np.repeat(spm_y[cur_row] - 1, x_sig_contribs.size),
# color=color_map.colors[1], lw=2)
# print info
print(activity)
print('Euler Diff at Max: {:.2f}'.format(ht_euler_add_max_mean))
print('Contrib Diff at Max: {:.2f}'.format(ht_contrib_max_mean))
print('HT Elevation significance:')
print(extract_sig(ht_euler_diff_vs_euler_add, x))
print(output_spm_p(ht_euler_diff_vs_euler_add))
dict(color=color_map.colors[0], marker=markers[0], markevery=20))
old_shr_ln_gt45 = mean_sd_plot(
axs[cur_row, 0], x, old_shr_mean_gt45, old_shr_sd_gt45,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[1], markevery=20))
new_shr_ln_lt35 = mean_sd_plot(
axs[cur_row, 1], x, new_shr_mean_lt35, new_shr_sd_lt35,
dict(color=color_map.colors[0], alpha=0.25),
dict(color=color_map.colors[0], marker=markers[0], markevery=20))
new_shr_ln_gt45 = mean_sd_plot(
axs[cur_row, 1], x, new_shr_mean_gt45, new_shr_sd_gt45,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[1], markevery=20))
# plot spm
x_old_sig = sig_filter(spm_old, x)
x_new_sig = sig_filter(spm_new, x)
axs[cur_row, 0].plot(x_old_sig,
np.repeat(12, x_old_sig.size),
color='k',
lw=2)
axs[cur_row, 1].plot(x_new_sig,
np.repeat(12, x_new_sig.size),
color='k',
lw=2)
if idx == 0:
leg_mean.append(old_shr_ln_lt35[0])
leg_mean.append(old_shr_ln_gt45[0])
leg_mean.append(new_shr_ln_lt35[0])
leg_mean.append(new_shr_ln_gt45[0])
'BP Min: {:.2f}, 25: {:.2f}, Median: {:.2f}, 75: {:.2f}, Max: {:.2f}'
.format(*retrieve_bp_stats(bp)))
print('BP Deg Median: {:.2f}, 25: {:.2f}, 75: {:.2f}, Max: {:.2f}'.
format(np.median(st_end_deg),
*np.quantile(st_end_deg, [0.25, 0.75]),
np.max(st_end_deg)))
print('GH Mean Deg: {:.2f} and Percent: {:.2f}'.format(
np.mean(gh_end_deg), np.mean(gh_percent)))
print('ST Mean Deg: {:.2f} and Percent: {:.2f}'.format(
np.mean(st_end_deg), np.mean(st_percent)))
st_vs_ht = spm_test(traj_st[:, 40:] / traj_ht[:, 40:],
0.10).inference(alpha,
two_tailed=True,
**infer_params)
x_gh_sig = sig_filter(st_vs_ht, x_er[40:])
axs_er[idx_act].plot(x_gh_sig, np.repeat(5, x_gh_sig.size), color='k')
print(extract_sig(st_vs_ht, x_er[40:]))
print('P-values: ')
print(output_spm_p(st_vs_ht))
# add stats test text
axs_elev[0].text(25,
10,
'GH ~ HT',
ha='left',
fontsize=10,
fontweight='bold')
axs_elev[0].text(107,
10,
'ST ~ HT',
cur_row = act_row[activity.lower()]
st_m_ln = mean_sd_plot(axs_elev[cur_row], x_elev, st_m_mean, st_m_sd,
dict(color=color_map.colors[0], alpha=0.25),
dict(color=color_map.colors[0], marker=markers[0], markevery=20))
st_f_ln = mean_sd_plot(axs_elev[cur_row], x_elev, st_f_mean, st_f_sd,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[1], markevery=20))
axs_elev[cur_row].errorbar(max_pos, st_m_max_mean, yerr=st_m_max_sd, color=color_map.colors[0],
marker=markers[0], capsize=3)
axs_elev[cur_row].errorbar(max_pos, st_f_max_mean, yerr=st_f_max_sd, color=color_map.colors[1],
marker=markers[1], capsize=3)
# spm
m_vs_f = spm_test(traj_st_m, traj_st_f).inference(alpha, two_tailed=True, **infer_params)
x_sig = sig_filter(m_vs_f, x_elev)
axs_elev[cur_row].plot(x_sig, np.repeat(spm_y[activity.lower()], x_sig.size), color='k')
print(activity)
print(extract_sig(m_vs_f, x_elev))
if idx == 0:
leg_elev_mean.append(st_m_ln[0])
leg_elev_mean.append(st_f_ln[0])
for idx_act, (activity, activity_df) in enumerate(db_er_endpts.groupby('Activity', observed=True)):
m_df = activity_df[activity_df['Gender'] == 'M']
f_df = activity_df[activity_df['Gender'] == 'F']
traj_st_m = np.stack(m_df['st_contribs_interp'], axis=0)[:, :, 2]
traj_st_f = np.stack(f_df['st_contribs_interp'], axis=0)[:, :, 2]
**infer_params)
latmed_vs_tilt = spm_test(traj_st_x - traj_st_z,
0).inference(alpha,
two_tailed=True,
**infer_params)
else:
latmed_vs_repro = spm_test(-(traj_st_x - traj_st_y),
0).inference(alpha,
two_tailed=True,
**infer_params)
latmed_vs_tilt = spm_test(-(traj_st_x - traj_st_z),
0).inference(alpha,
two_tailed=True,
**infer_params)
x_sig_repro = sig_filter(latmed_vs_repro, x_elev)
x_sig_tilt = sig_filter(latmed_vs_tilt, x_elev)
axs_elev[cur_row].plot(x_sig_repro,
np.repeat(spm_y[activity.lower()],
x_sig_repro.size),
'-',
color='k')
axs_elev[cur_row].plot(x_sig_tilt,
np.repeat(spm_y[activity.lower()] - 3,
x_sig_tilt.size),
'-',
color='k')
print('Activity: {}'.format(activity))
print('Repro Sig')
print(extract_sig(latmed_vs_repro, x_elev))
dict(color=color_map.colors[0], marker=markers[0], markevery=20))
st_m_ln = mean_sd_plot(axs[idx_act, 1], x, st_m_mean, st_m_sd,
dict(color=color_map.colors[1], alpha=0.2),
dict(color=color_map.colors[1], marker=markers[1], markevery=20))
gh_f_ln = mean_sd_plot(axs[idx_act, 2], x, gh_f_mean, gh_f_sd,
dict(color=color_map.colors[0], alpha=0.2),
dict(color=color_map.colors[0], marker=markers[0], markevery=20))
gh_m_ln = mean_sd_plot(axs[idx_act, 2], x, gh_m_mean, gh_m_sd,
dict(color=color_map.colors[1], alpha=0.2),
dict(color=color_map.colors[1], marker=markers[1], markevery=20))
ht_f_vs_m = spm_test(traj_ht_f[:, 1:], traj_ht_m[:, 1:]).inference(alpha, two_tailed=True, **infer_params)
st_f_vs_m = spm_test(traj_st_f[:, 1:], traj_st_m[:, 1:]).inference(alpha, two_tailed=True, **infer_params)
gh_f_vs_m = spm_test(traj_gh_f[:, 1:], traj_gh_m[:, 1:]).inference(alpha, two_tailed=True, **infer_params)
ht_x_sig = sig_filter(ht_f_vs_m, x[1:])
st_x_sig = sig_filter(st_f_vs_m, x[1:])
gh_x_sig = sig_filter(gh_f_vs_m, x[1:])
axs[idx_act, 0].plot(ht_x_sig, np.repeat(5, ht_x_sig.size), color='k', lw=2)
axs[idx_act, 1].plot(st_x_sig, np.repeat(5, st_x_sig.size), color='k', lw=2)
axs[idx_act, 2].plot(gh_x_sig, np.repeat(5, gh_x_sig.size), color='k', lw=2)
if idx_act == 0:
# legend lines
mean_lns.append(ht_f_ln[0])
mean_lns.append(ht_m_ln[0])
mean_lns.append(st_f_ln[0])
mean_lns.append(st_m_ln[0])
mean_lns.append(gh_f_ln[0])
mean_lns.append(gh_m_ln[0])
repro_ln_lt35 = mean_sd_plot(axs[cur_row, 0], x, repro_mean_lt35, repro_sd_lt35,
dict(color=color_map.colors[0], alpha=0.25),
dict(color=color_map.colors[0], marker=markers[0], markevery=20))
poe_ln_lt35 = mean_sd_plot(axs[cur_row, 1], x, poe_mean_lt35, poe_sd_lt35,
dict(color=color_map.colors[0], alpha=0.25),
dict(color=color_map.colors[0], marker=markers[2], markevery=20))
repro_ln_gt45 = mean_sd_plot(axs[cur_row, 0], x, repro_mean_gt45, repro_sd_gt45,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[0], markevery=20))
poe_ln_gt45 = mean_sd_plot(axs[cur_row, 1], x, poe_mean_gt45, poe_sd_gt45,
dict(color=color_map.colors[1], alpha=0.25),
dict(color=color_map.colors[1], marker=markers[2], markevery=20))
# plot SPM
repro_x_sig = sig_filter(repro_lt35_vs_gt35, x)
poe_x_sig = sig_filter(poe_lt35_vs_gt35, x)
axs[cur_row, 0].plot(repro_x_sig, np.repeat(ax_limits[0][1], repro_x_sig.size), color='k', lw=2)
axs[cur_row, 1].plot(poe_x_sig, np.repeat(ax_limits[1][1], poe_x_sig.size), color='k', lw=2)
if act_idx == 0:
leg_mean.append(repro_ln_lt35[0])
leg_mean.append(repro_ln_gt45[0])
leg_mean.append(poe_ln_lt35[0])
leg_mean.append(poe_ln_gt45[0])
# figure title and legend
plt.figure(fig.number)
plt.tight_layout(pad=0.25, h_pad=1.5, w_pad=0)
fig.suptitle('ST ReProtraction and GH PoE Comparison by Age Group', x=0.5, y=0.99, fontweight='bold')
plt.subplots_adjust(top=0.93)
