def corr(source=False, make_tight=True, force_new=False):
all_timecourses = get_et_data(make='timecourse', make_categories=categories, savefile='time_series.csv', force_new=force_new)
timecourse_means = all_timecourses.groupby(level=(1,2)).mean()
pearson = pearsonr(timecourse_means["L Dia X [px]"], timecourse_means["L Dia Y [px]"])
fig = figure(figsize=(3, 3), dpi=300,facecolor='#eeeeee', tight_layout=make_tight)
ax1=fig.add_subplot(1,1,1)
ax1.yaxis.grid(True, linestyle='-', which='major', color='#dddddd',alpha=0.6, zorder = 0)
ax1.set_axisbelow(True)
ax1.plot(all_timecourses["L Dia X [px]"][::3], all_timecourses["L Dia Y [px]"][::3],".",markersize=2, color='0.91',zorder=1)
all_points = Rectangle((0, 0), 1, 1, color="0.9")
ax1.plot(timecourse_means["L Dia X [px]"], timecourse_means["L Dia Y [px]"],".",markersize=5, alpha=0.7, markerfacecolor='#ffffff', markeredgecolor='g',zorder=3)
participant_means = Rectangle((0, 0), 1, 1, ec="g", fc="#ffffff")
A = np.vstack([timecourse_means["L Dia X [px]"], np.ones(len(timecourse_means["L Dia X [px]"]))]).T
m, c = np.linalg.lstsq(A, timecourse_means["L Dia Y [px]"])[0]
ax1.plot(np.array(all_timecourses["L Dia X [px]"])[10000:],np.array(all_timecourses["L Dia X [px]"])[10000:]*m+c,color='m', linewidth=0.8,antialiased=True,zorder=2)
regression = Rectangle((0, 0), 1, 1, color="m")
ax1.tick_params(axis='both', labelsize=8)
ax1.set_ylabel('Y-axis Pupil Diameter [px]', fontsize=9)
ax1.set_xlabel('X-axis Pupil Diameter [px]', fontsize=9)
ax1.set_ylim(bottom=10)
ax1.set_xlim(left=10)
legend((all_points, participant_means, regression),('Raw','Per-(Time, Condition) Particiant Means', 'Regression'),loc='upper center', bbox_to_anchor=(0.5, 1.038), ncol=3, fancybox=False, shadow=False, prop=FontProperties(size='7'))
return [list(pearson), m, c]
def main(make=False, source=False, make_tight=True, compare="difficulty", show="", make_std=True, make_sem=True, ecolor='0.3', elinewidth=2, total="means", make_scrambled_yn=False, per_participant_error=False, fontscale=1, force_new=False):
data_all = get_et_data(make='timecourse', make_categories=categories_of_interest, savefile='time_series_CoI.csv', baseline="participant", force_new=force_new)
data_all["Time"] = data_all["Time"]/1000 #make seconds (from milliseconds)
data_all = data_all.reset_index(drop=False)
data_all = data_all.set_index(["CoI","measurement","ID"], append=True, drop=True)
data_all = data_all.reset_index(level=0,drop=True)
ids = sorted(list(data_all.index.levels[2]))
pos_ids = np.arange(len(ids))
fig = figure(figsize=(pos_ids.max()*5, 4), dpi=300,facecolor='#eeeeee', tight_layout=make_tight)
ax=fig.add_subplot(1,1,1)
ax.yaxis.grid(True, linestyle='-', which='major', color='#dddddd',alpha=0.6, zorder = 0)
ax.set_axisbelow(True)
width = 0.12
#below this: per-participant graphs
fix = plt.bar(pos_ids-3*width, data_all.ix["fix"].groupby(level=(1))['Pupil'].mean(), width ,color='k', alpha=0.25, zorder = 1, linewidth=0)
plot_em_easy = plt.bar(pos_ids-2*width, data_all.ix["emotion-easy"].groupby(level=(1))['Pupil'].mean(), width ,color='m', alpha=0.7, zorder = 1, linewidth=0)
plot_em_hard = plt.bar(pos_ids-width, data_all.ix["emotion-hard"].groupby(level=(1))['Pupil'].mean(), width ,color='m', alpha=0.4, zorder = 1, linewidth=0)
plot_sc_easy = plt.bar(pos_ids, data_all.ix["scrambling-easy"].groupby(level=(1))['Pupil'].mean(), width ,color='g', alpha=0.7, zorder = 1, linewidth=0)
plot_sc_hard = plt.bar(pos_ids+width, data_all.ix["scrambling-hard"].groupby(level=(1))['Pupil'].mean(), width ,color='g', alpha=0.4, zorder = 1, linewidth=0)
if per_participant_error:
if make_std:
for idx, category in enumerate([category_data[1] for category_data in categories_of_interest]):
errorbar(pos_ids+(width*(-2.5+1*idx)), data_all.ix[category].groupby(level=(1))['Pupil'].mean(), yerr=data_all.ix[category].groupby(level=1)['Pupil'].aggregate(np.std), ecolor=str(float(ecolor)+0.25), elinewidth=elinewidth, capsize=0, linestyle='None', zorder = 2)
if make_sem:
for idx, category in enumerate([category_data[1] for category_data in categories_of_interest]):
errorbar(pos_ids+(width*(-2.5+1*idx)), data_all.ix[category].groupby(level=(1))['Pupil'].mean(), yerr=data_all.ix[category].groupby(level=1)['Pupil'].aggregate(sem), ecolor=ecolor, elinewidth=elinewidth, capsize=0, linestyle='None', zorder = 2)
#below this: total graphs
plt.bar(pos_ids[-1]+1-width, data_all.ix["fix"]['Pupil'].mean(), width ,color='k', alpha=0.25, zorder = 1, linewidth=0)
plt.bar(pos_ids[-1]+1, data_all.ix["emotion-easy"]['Pupil'].mean(), width ,color='m', alpha=0.7, zorder = 1, linewidth=0)
plt.bar(pos_ids[-1]+1+width, data_all.ix["emotion-hard"]['Pupil'].mean(), width ,color='m', alpha=0.4, zorder = 1, linewidth=0)
plt.bar(pos_ids[-1]+1+2*width, data_all.ix["scrambling-easy"]['Pupil'].mean(), width ,color='g', alpha=0.7, zorder = 1, linewidth=0)
plt.bar(pos_ids[-1]+1+3*width, data_all.ix["scrambling-hard"]['Pupil'].mean(), width ,color='g', alpha=0.4, zorder = 1, linewidth=0)
if total == 'all':
if make_std:
for idx, category in enumerate([category_data[1] for category_data in categories_of_interest]):
errorbar(pos_ids[-1]+1+(width*(-0.5+1*idx)), data_all.ix[category]['Pupil'].mean(), yerr=np.std(data_all.ix[category]['Pupil']), ecolor=str(float(ecolor)+0.25), elinewidth=elinewidth, capsize=0, linestyle='None', zorder = 2)
if make_sem:
for idx, category in enumerate([category_data[1] for category_data in categories_of_interest]):
errorbar(pos_ids[-1]+1+(width*(-0.5+1*idx)), data_all.ix[category]['Pupil'].mean(), yerr=sem(data_all.ix[category]['Pupil']), ecolor=ecolor, elinewidth=elinewidth, capsize=0, linestyle='None', zorder = 2)
elif total == 'means':
if make_std:
for idx, category in enumerate([category_data[1] for category_data in categories_of_interest]):
errorbar(pos_ids[-1]+1+(width*(-0.5+1*idx)), data_all.ix[category]['Pupil'].mean(), yerr=np.std(data_all.ix[category].groupby(level=1)['Pupil'].mean()), ecolor=str(float(ecolor)+0.25), elinewidth=elinewidth, capsize=0, linestyle='None', zorder = 2)
if make_sem:
for idx, category in enumerate([category_data[1] for category_data in categories_of_interest]):
errorbar(pos_ids[-1]+1+(width*(-0.5+1*idx)), data_all.ix[category]['Pupil'].mean(), yerr=sem(data_all.ix[category].groupby(level=1)['Pupil'].mean()), ecolor=ecolor, elinewidth=elinewidth, capsize=0, linestyle='None', zorder = 2)
width_multiplier = 4
plt.axvline(pos_ids[-1]+1-width*width_multiplier, color='0.2') #separator - per-person/total
ids=ids+[' ALL']
pos_ids = np.arange(len(ids))
ax.set_ylabel('Pupil Area Ratio', fontsize=11*fontscale)
ax.set_xlabel('Participant', fontsize=11*fontscale)
ax.set_xticks(pos_ids)
ax.set_xticklabels(ids,fontsize=7.5*fontscale,rotation=30) # add rotation=30 if things get too crowded
for tick in ax.axes.get_xticklines():
tick.set_visible(False)
ax.set_xlim(0, pos_ids[-1]+width*1) # before scaling to add padding in front of zero
axis.Axis.zoom(ax.xaxis, -0.8) # sets x margins further apart from the content proportional to its length
axis.Axis.zoom(ax.yaxis, -0.5) # sets y margins further apart from the content proportional to its length
ax.set_ylim(bottom=0) # after scaling to disregard padding unerneath zero.
legend((fix, plot_em_easy, plot_em_hard, plot_sc_easy, plot_sc_hard),("Fixation",'Strong Emotion','Weak Emotion', "Easy Scrambling", "Hard Scrambling"),loc='upper center', bbox_to_anchor=(0.5, 1.05), ncol=3, fancybox=False, shadow=False, prop=FontProperties(size=str(8*fontscale)))
return data_all
def time_course(source=False, make_tight=True, make_sem=True, show=["emotion", "scrambled", "rt_em", "rt_sc"], sample_by=4, fontscale=1, force_new=False):
rt = get_rt_data(make_categories=categories)
all_timecourses = get_et_data(make='timecourse', make_categories=categories, savefile='time_series.csv', baseline="participant", force_new=force_new)
all_timecourses["Time"] = all_timecourses["Time"]/1000 #make seconds (from milliseconds)
all_timecourses.to_csv("/home/chymera/TC_all.csv")
all_timecourses = all_timecourses.reset_index(drop=False)
all_timecourses = all_timecourses.set_index(["CoI","measurement"], append=True, drop=True)
all_timecourses = all_timecourses.reset_index(level=0,drop=True)
#downsamples the timecourse:
timecourse_plot = all_timecourses.groupby(level=0).apply(downsample, sample=sample_by, group='measurement')
#BEGI SEMS
SEM_timecourse_normed = all_timecourses.groupby(level=(0,1)).aggregate(sem)
SEM_timecourse_plot = SEM_timecourse_normed.groupby(level=0).apply(downsample, sample=sample_by, group='measurement')
###END SEMS
timecourse_plot = timecourse_plot.reset_index(level=1)
SEM_timecourse_plot = SEM_timecourse_plot.reset_index(level=1)
#BEGIN PLOTTING
fig = figure(figsize=(5, 3), dpi=300,facecolor='#eeeeee', tight_layout=make_tight)
ax1=fig.add_subplot(1,1,1,axisbg='1')
ax1.yaxis.grid(True, linestyle='-', which='major', color='#dddddd',alpha=0.6, zorder = 0)
ax1.set_axisbelow(True)
plotted = []
plotted_names = []
if "fix" in show:
if make_sem:
ax1.fill_between(np.array(timecourse_plot.ix["fix"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["fix"]["Pupil"]+SEM_timecourse_plot.ix["fix"]["Pupil"]/2), np.array(timecourse_plot.ix["fix"]["Pupil"]-SEM_timecourse_plot.ix["fix"]["Pupil"]/2), facecolor="0.8", edgecolor="none", alpha=0.2, zorder=0)
ax1.plot(np.array(timecourse_plot.ix["fix"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["fix"]["Pupil"]), color="0.8",zorder=2)
fix = Rectangle((0, 0), 1, 1, color="0.8")
plotted.append(fix)
plotted_names.append("Fixation")
if "easy" in show:
if make_sem:
ax1.fill_between(np.array(timecourse_plot.ix["easy"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["easy"]["Pupil"]+SEM_timecourse_plot.ix["easy"]["Pupil"]/2), np.array(timecourse_plot.ix["easy"]["Pupil"]-SEM_timecourse_plot.ix["easy"]["Pupil"]/2), facecolor="g", edgecolor="none", alpha=0.1, zorder=0)
ax1.plot(np.array(timecourse_plot.ix["easy"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["easy"]["Pupil"]), color='g',zorder=2)
easy = Rectangle((0, 0), 1, 1, color="g")
plotted.append(easy)
plotted_names.append("Easy Trials")
if "hard" in show:
if make_sem:
ax1.fill_between(np.array(timecourse_plot.ix["hard"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["hard"]["Pupil"]+SEM_timecourse_plot.ix["hard"]["Pupil"]/2), np.array(timecourse_plot.ix["hard"]["Pupil"]-SEM_timecourse_plot.ix["hard"]["Pupil"]/2), facecolor="m", edgecolor="none", alpha=0.1, zorder=0)
ax1.plot(np.array(timecourse_plot.ix["hard"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["hard"]["Pupil"]), color='m',zorder=2)
hard = Rectangle((0, 0), 1, 1, color="m")
plotted.append(hard)
plotted_names.append("Hard Trials")
if "happy" in show:
if make_sem:
ax1.fill_between(np.array(timecourse_plot.ix["happy"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["happy"]["Pupil"]+SEM_timecourse_plot.ix["happy"]["Pupil"]/2), np.array(timecourse_plot.ix["happy"]["Pupil"]-SEM_timecourse_plot.ix["happy"]["Pupil"]/2), facecolor="g", edgecolor="none", alpha=0.1, zorder=0)
ax1.plot(np.array(timecourse_plot.ix["happy"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["happy"]["Pupil"]), color='g',zorder=2)
happy = Rectangle((0, 0), 1, 1, color="g")
plotted.append(happy)
plotted_names.append("'Happy' Trials")
if "fearful" in show:
if make_sem:
ax1.fill_between(np.array(timecourse_plot.ix["fearful"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["fearful"]["Pupil"]+SEM_timecourse_plot.ix["fearful"]["Pupil"]/2), np.array(timecourse_plot.ix["fearful"]["Pupil"]-SEM_timecourse_plot.ix["fearful"]["Pupil"]/2), facecolor="m", edgecolor="none", alpha=0.1, zorder=0)
ax1.plot(np.array(timecourse_plot.ix["fearful"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["fearful"]["Pupil"]), color='m',zorder=2)
fearful = Rectangle((0, 0), 1, 1, color="m")
plotted.append(fearful)
plotted_names.append("'Fearful' Trials")
if "emotion" in show:
tc = (np.array(timecourse_plot.ix["fearful"]["measurement"])/(60/sample_by)+np.array(timecourse_plot.ix["happy"]["measurement"])/(60/sample_by))/2
v = (np.array(timecourse_plot.ix["fearful"]["Pupil"])+np.array(timecourse_plot.ix["happy"]["Pupil"]))/2
if make_sem:
se = (np.array(SEM_timecourse_plot.ix["fearful"]["Pupil"])+np.array(SEM_timecourse_plot.ix["happy"]["Pupil"]))/4
ax1.fill_between(tc, v+se, v-se, facecolor="g", edgecolor="none", alpha=0.1, zorder=0)
ax1.plot(tc, v, color='g',zorder=2)
emotion = Rectangle((0, 0), 1, 1, color="g")
plotted.append(emotion)
plotted_names.append("Emotion Trials")
if "scrambled" in show:
if make_sem:
ax1.fill_between(np.array(timecourse_plot.ix["scrambled"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["scrambled"]["Pupil"]+SEM_timecourse_plot.ix["scrambled"]["Pupil"]/2), np.array(timecourse_plot.ix["scrambled"]["Pupil"]-SEM_timecourse_plot.ix["scrambled"]["Pupil"]/2), facecolor="m", edgecolor="none", alpha=0.1, zorder=0)
ax1.plot(np.array(timecourse_plot.ix["scrambled"]["measurement"])/(60/sample_by), np.array(timecourse_plot.ix["scrambled"]["Pupil"]), color='m',zorder=2)
scrambled = Rectangle((0, 0), 1, 1, color="m")
plotted.append(scrambled)
plotted_names.append("Scrambled Trials")
if "all" in show:
all_tc = (np.array(timecourse_plot.ix["easy"]["measurement"])/(60/sample_by)+np.array(timecourse_plot.ix["hard"]["measurement"])/(60/sample_by))/2
all_v = (np.array(timecourse_plot.ix["easy"]["Pupil"])+np.array(timecourse_plot.ix["hard"]["Pupil"]))/2
if make_sem:
all_se = (np.array(SEM_timecourse_plot.ix["easy"]["Pupil"])+np.array(SEM_timecourse_plot.ix["hard"]["Pupil"]))/4
ax1.fill_between(all_tc, all_v+all_se, all_v-all_se, facecolor="g", edgecolor="none", alpha=0.1, zorder=0)
ax1.plot(all_tc, all_v, color='g',zorder=2)
ALL = Rectangle((0, 0), 1, 1, color="g")
plotted.append(ALL)
plotted_names.append("All Trials")
if "rt_e" in show:
ax1.axvline(rt[(rt["difficulty"] == "easy")]["RT"].mean(), linewidth=0.3, color='g', zorder=1)
if "rt_h" in show:
ax1.axvline(rt[(rt["difficulty"] == "hard")]["RT"].mean(), linewidth=0.3, color='m', zorder=1)
if "rt_ha" in show:
ax1.axvline(rt[(rt["emotion"] == "happy")]["RT"].mean(), linewidth=0.3, color='g', zorder=1)
if "rt_fe" in show:
ax1.axvline(rt[(rt["emotion"] == "fearful")]["RT"].mean(), linewidth=0.3, color='m', zorder=1)
if "rt_em" in show:
ax1.axvline(rt[(rt["emotion"] != "scrambled")]["RT"].mean(), linewidth=0.3, color='g', zorder=1)
if "rt_sc" in show:
ax1.axvline(rt[(rt["emotion"] == "scrambled")]["RT"].mean(), linewidth=0.3, color='m', zorder=1)
if "rt_all" in show:
ax1.axvline(rt[(rt["difficulty"] == "easy") | (rt["difficulty"] == "hard")]["RT"].mean(), linewidth=0.3, color='g', zorder=1)
ax1.tick_params(axis='both', labelsize=8*fontscale)
ax1.set_ylabel('Pupil Area Ratio', fontsize=11*fontscale)
ax1.set_xlabel('Time [s]', fontsize=11*fontscale)
legend((plotted),(plotted_names),loc='upper center', bbox_to_anchor=(0.5, 1.06), ncol=3, fancybox=False, shadow=False, prop=FontProperties(size=str(9*fontscale)))
#END PLOTTING
return all_timecourses
def discrete_time(make_tight=True, show="", sample_size=40, fontscale=1):
df = get_et_data(make='timecourse', make_categories=categories, savefile='time_series.csv', baseline="participant", force_new=False)
df["Time"] = df["Time"]/1000 #make seconds (from milliseconds)
discretize = {0:"S01", 1:"S02", 2:"S03", 3:"S04", 4:"S05", 5:"S06", 6:"S07", 7:"S08", 8:"S09", 9:"S10", 10:"S11", 11:"S12", 12:"S13", 13:"S14", 14:"S15", 15:"S16", 16:"S17", 17:"S18", 18:"S19", 19:"S20"}
df = df.groupby(level=[0,1]).apply(downsample, sample=sample_size+1, group='measurement')
df.reset_index(inplace=True)
#fixation has a jitter and building a mean gives smaller time values for the last time points:
df.ix[(df["CoI"] == "fix"), "Time"] = (df[(df["CoI"] == "easy")]["Time"]+df[(df["CoI"] == "hard")]["Time"])/2
df["dTime"] = df["measurement"].copy()
df["dTime"] = df["dTime"].apply(lambda x: discretize.get(x,x))
df_means = df.copy()
df_means = df_means.set_index(['ID',"CoI", "measurement"], append=False, drop=True)
df_means = df_means.groupby(level=(1,2)).mean()
#BEGIN PLOTTING
fig = figure(figsize=(5, 3), dpi=300,facecolor='#eeeeee', tight_layout=make_tight)
ax1=fig.add_subplot(1,1,1,axisbg='1')
ax1.yaxis.grid(True, linestyle='-', which='major', color='#dddddd',alpha=0.6, zorder = 0)
ax1.set_axisbelow(True)
ids = sorted(list(set(list(df["dTime"]))))
pos_ids = np.arange(len(ids))
plotted = []
plotted_names = []
if "easy" in show:
ax1.plot(pos_ids, np.array(df_means.ix["easy"]["Pupil"]), "^-", markeredgecolor="none", linewidth=0.2, color='g')
easy = Rectangle((0, 0), 1, 1, color="g")
plotted.append(easy)
plotted_names.append("Easy Trials")
if "hard" in show:
ax1.plot(pos_ids, np.array(df_means.ix["hard"]["Pupil"]), "v-", markeredgecolor="none", linewidth=0.2, color='m')
hard = Rectangle((0, 0), 1, 1, color="m")
plotted.append(hard)
plotted_names.append("Hard Trials")
if "happy" in show:
ax1.plot(pos_ids, np.array(df_means.ix["happy"]["Pupil"]), "^-", markeredgecolor="none", linewidth=0.2, color='g')
happy = Rectangle((0, 0), 1, 1, color="g")
plotted.append(happy)
plotted_names.append("Happy Trials")
if "fearful" in show:
ax1.plot(pos_ids, np.array(df_means.ix["fearful"]["Pupil"]), "v-", markeredgecolor="none", linewidth=0.2, color='m')
fearful = Rectangle((0, 0), 1, 1, color="m")
plotted.append(fearful)
plotted_names.append("Fearful Trials")
if "emotion" in show:
ax1.plot(pos_ids, (np.array(df_means.ix["happy"]["Pupil"])+np.array(df_means.ix["fearful"]["Pupil"]))/2, "^-", markeredgecolor="none", linewidth=0.2, color='g')
emotion = Rectangle((0, 0), 1, 1, color="g")
plotted.append(emotion)
plotted_names.append("Emotion Trials")
if "scrambled" in show:
ax1.plot(pos_ids, np.array(df_means.ix["scrambled"]["Pupil"]), "v-", markeredgecolor="none", linewidth=0.2, color='m')
scrambled = Rectangle((0, 0), 1, 1, color="m")
plotted.append(scrambled)
plotted_names.append("Scrambled Trials")
ax1.set_xticks(pos_ids)
ax1.set_xticklabels(ids,fontsize=9*fontscale) # add rotation=30 if things get too crowded
ax1.set_ylabel('Pupil Area Ratio', fontsize=11*fontscale)
ax1.set_xlabel('Time Series Sextiles [66 ms]', fontsize=11*fontscale)
axis.Axis.zoom(ax1.xaxis, -0.6) # sets x margins further apart from the content proportional to its length
axis.Axis.zoom(ax1.yaxis, -0.4) # sets x margins further apart from the content proportional to its length
legend((plotted),(plotted_names),loc='upper center', bbox_to_anchor=(0.5, 1.06), ncol=3, fancybox=False, shadow=False, prop=FontProperties(size=str(5*fontscale)))
#END PLOTTING
return df
