def thresholds(save_pdf=False):
conf = ss_timing_analysis.conf.get_conf()
# load the fit parameters, excluding the bad subjects
# this will be subj x onsets x oris x (a,b) x (est, 2.5, 97.5)
(fit, _, _) = ss_timing_analysis.group_fit.load_fit_data(exclude=True)
# restrict to just the alpha estimates
data = fit[..., 0, 0]
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
page = embed.Root.Add("page")
page.width.val = "13.5cm"
page.height.val = "6cm"
# separate columns for simultaneous and leading
grid = page.Add("grid")
grid.rows.val = 1
grid.columns.val = 2
grid.leftMargin.val = grid.rightMargin.val = "0cm"
grid.topMargin.val = grid.bottomMargin.val = "0cm"
onset_order = [1, 0] # sim, leading
ori_order = [0, 1] # orth, para
for i_onset in onset_order:
graph = grid.Add("graph", autoadd=False)
graph.bottomMargin.val = "1cm"
graph.topMargin.val = "0.6cm"
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
for i_ori in ori_order:
curr_data = data[:, i_onset, i_ori]
boxplot = graph.Add("boxplot")
dataset_str = "data_{onset:d}_{ori:d}".format(onset=i_onset,
ori=i_ori)
embed.SetData(dataset_str, curr_data)
boxplot.values.val = dataset_str
boxplot.posn.val = i_ori
boxplot.labels.val = conf.surr_ori_labels[i_ori]
boxplot.fillfraction.val = 0.3
boxplot.markerSize.val = "2pt"
for i_subj in xrange(conf.n_subj):
subj_data_str = "subj_data_{subj:d}_{onset:d}".format(
subj=i_subj, onset=i_onset)
embed.SetData(subj_data_str, data[i_subj, i_onset, :])
xy = graph.Add("xy")
xy.xData.val = [0, 1]
xy.yData.val = subj_data_str
xy.MarkerFill.hide.val = True
xy.MarkerLine.hide.val = True
xy.PlotLine.transparency.val = 80
x_axis.mode.val = "labels"
x_axis.MajorTicks.manualTicks.val = [0, 1]
x_axis.MinorTicks.hide.val = True
x_axis.label.val = "Relative orientation"
y_axis.log.val = True
y_axis.TickLabels.format.val = "%.3g"
y_axis.MajorTicks.manualTicks.val = [0.001, 0.01, 0.1, 0.5, 1]
y_axis.min.val = 0.005
y_axis.max.val = 1.0
y_axis.label.val = "Threshold contrast"
cond_label = graph.Add("label")
cond_label.label.val = conf.surr_onset_labels[i_onset]
cond_label.yPos.val = 1.02
cond_label.xPos.val = 0.5
cond_label.alignHorz.val = "centre"
cond_label.Text.size.val = "8pt"
if save_pdf:
_save(embed, conf, "ss_timing_thresholds")
embed.EnableToolbar(True)
embed.WaitForClose()
def eg_subject(subj_id="p1022", save_pdf=False):
conf = ss_timing_analysis.conf.get_conf()
# find the index for this subject
# we'll load the data without excluding participants, so this needs to be
# in the all subjects list
i_subj = conf.all_subj_ids.index(subj_id)
# subj x onsets x oris x bins x (prop, n)
data = ss_timing_analysis.group_data.bin_group_data()
# restrict to this subject
data = data[i_subj, ...]
# fit is: subj x onsets x oris x (a, b) x (est, 2.5, 97.5)
# fit_fine is: subj x onsets x oris x X x 2 (2.5, 97.5)
(fit, fit_fine, _) = ss_timing_analysis.group_fit.load_fit_data()
# again, restrict to this subject
fit = fit[i_subj, ...]
fit_fine = fit_fine[i_subj, ...]
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
embed.SetData("bin_centres", conf.bin_centres)
embed.SetData("fine_x", conf.fine_x)
page = embed.Root.Add("page")
page.width.val = "8cm"
page.height.val = "20cm"
grid = page.Add("grid")
grid.rows.val = 4
grid.columns.val = 1
grid.leftMargin.val = grid.rightMargin.val = "0cm"
grid.topMargin.val = grid.bottomMargin.val = "0cm"
onset_order = [1, 0] # sim, leading
ori_order = [0, 1] # orth, par
for i_onset in onset_order:
for i_ori in ori_order:
graph = grid.Add("graph", autoadd=False)
graph.leftMargin.val = "1.2cm"
graph.rightMargin.val = "0.5cm"
graph.topMargin.val = "0.7cm"
graph.bottomMargin.val = "0.85cm"
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
cond_label = graph.Add("label")
cond_label.label.val = ", ".join(
[conf.surr_ori_labels[i_ori], conf.surr_onset_labels[i_onset]])
cond_label.yPos.val = 1.02
cond_label.xPos.val = 0.5
cond_label.alignHorz.val = "centre"
cond_label.Text.size.val = "8pt"
# CROSSHAIRS
pse_y = graph.Add("xy")
pse_y.xData.val = [0.001, fit[i_onset, i_ori, 0, 0]]
pse_y.yData.val = [
conf.psych_func(fit[i_onset, i_ori, 0, 0],
fit[i_onset, i_ori, 0, 0], fit[i_onset, i_ori,
0, 1])
] * 2
pse_x = graph.Add("xy")
pse_x.xData.val = [fit[i_onset, i_ori, 0, 0]] * 2
pse_x.yData.val = [
-0.05,
conf.psych_func(fit[i_onset, i_ori, 0, 0],
fit[i_onset, i_ori, 0, 0], fit[i_onset, i_ori,
0, 1])
]
for pse_ax in (pse_y, pse_x):
pse_ax.MarkerFill.hide.val = True
pse_ax.MarkerLine.hide.val = True
pse_ax.PlotLine.style.val = "dashed"
# POINTS
points = graph.Add("xy")
prop_name = "resp_prop_{t:d}_{o:d}".format(t=i_onset, o=i_ori)
embed.SetData(prop_name, data[i_onset, i_ori, :, 0])
k_name = "resp_k_{t:d}_{o:d}".format(t=i_onset, o=i_ori)
point_scale = np.sqrt(
(data[i_onset, i_ori, :, 1] * 1) / np.pi) * 2 * 0.35
embed.SetData(k_name, point_scale)
points.xData.val = "bin_centres"
points.yData.val = prop_name
points.scalePoints.val = k_name
points.MarkerLine.hide.val = True
points.MarkerFill.transparency.val = 65
points.PlotLine.hide.val = True
points.MarkerFill.color.val = "blue"
# FIT
fit_plot = graph.Add("xy")
fit_name = "fit_{t:d}_{o:d}".format(t=i_onset, o=i_ori)
fit_y = conf.psych_func(conf.fine_x,
alpha=fit[i_onset, i_ori, 0, 0],
beta=fit[i_onset, i_ori, 1, 0])
embed.SetData(
fit_name,
fit_y,
poserr=np.abs(fit_fine[i_onset, i_ori, :, 1] - fit_y),
negerr=np.abs(fit_fine[i_onset, i_ori, :, 0] - fit_y))
fit_plot.xData.val = "fine_x"
fit_plot.yData.val = fit_name
fit_plot.MarkerFill.hide.val = True
fit_plot.MarkerLine.hide.val = True
fit_plot.errorStyle.val = "fillvert"
fit_plot.ErrorBarLine.hide.val = True
x_axis.log.val = True
x_axis.label.val = "Target contrast"
x_axis.TickLabels.format.val = "%.3g"
x_axis.MajorTicks.manualTicks.val = [0.001, 0.01, 0.1, 0.5, 1]
y_axis.min.val = -0.1
y_axis.max.val = 1.1
y_axis.label.val = "Accuracy (prop. correct)"
y_axis.MinorTicks.hide.val = True
y_axis.MajorTicks.manualTicks.val = [0, 0.25, 0.5, 0.69, 1]
y_axis.TickLabels.format.val = "%.02f"
if save_pdf:
_save(embed, conf, "ss_timing_eg_subject")
embed.EnableToolbar(True)
embed.WaitForClose()
def scatter(cond, form, save_pdf=False):
if cond not in ["sim", "sim_orth", "lead"]:
raise ValueError()
if form not in ["linear", "rank"]:
raise ValueError()
conf = ss_timing_analysis.conf.get_conf()
# load the fit parameters, excluding the bad subjects
# this will be subj x onsets x oris x (a,b) x (est, 2.5, 97.5)
(fit, _, _) = ss_timing_analysis.group_fit.load_fit_data(exclude=True)
# restrict to just the alpha estimates
data = fit[..., 0, 0]
if cond == "sim":
data = data[:, 1, 1] - data[:, 1, 0]
elif cond == "sim_orth":
data = data[:, 1, 0]
elif cond == "lead":
#data = (data[:, 1, 1] - data[:, 1, 0]) - (data[:, 0, 1] - data[:, 0, 0])
data = data[:, 0, 1] - data[:, 0, 0]
if form == "rank":
data = scipy.stats.rankdata(data)
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
embed.SetData("data", data)
page = embed.Root.Add("page")
page.width.val = "13.5cm"
page.height.val = "10cm"
grid = page.Add("grid")
grid.rows.val = 2
grid.columns.val = 2
grid.leftMargin.val = grid.rightMargin.val = "2cm"
grid.topMargin.val = grid.bottomMargin.val = "0cm"
ss_nice = [
"Unusual experiences", "Cognitive disorganisation",
"Introvertive anhedonia", "Impulsive nonconformity"
]
for (i_sub, subscale) in enumerate(
("un_ex", "cog_dis", "int_anh", "imp_non")):
curr_ss = ss_timing_analysis.dem.get_olife_subscale(subscale,
exclude=True)
if form == "rank":
curr_ss = scipy.stats.rankdata(curr_ss)
assert len(curr_ss) == len(data)
graph = grid.Add("graph", autoadd=False)
graph.bottomMargin.val = "1cm"
graph.topMargin.val = "0.6cm"
graph.leftMargin.val = "1cm"
graph.aspect.val = 1
label = graph.Add("label")
label.label.val = ss_nice[i_sub]
label.yPos.val = 1.025
label.xPos.val = 0.5
label.alignHorz.val = "centre"
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
xy = graph.Add("xy")
xy.xData.val = curr_ss
xy.yData.val = data
xy.PlotLine.hide.val = True
xy.MarkerFill.transparency.val = 60
xy.MarkerLine.hide.val = True
xy.markerSize.val = "3pt"
if form == "rank":
x_axis.label.val = "O-LIFE subscale (rank; 1 = lowest)"
else:
x_axis.label.val = "O-LIFE subscale score"
if cond == "sim":
if form == "rank":
y_axis.label.val = "Context effect (rank; 1 = lowest)"
else:
y_axis.label.val = "Context effect (contrast units)"
elif cond == "sim_orth":
if form == "rank":
y_axis.label.val = "Orthogonal threshold (rank; 1 = lowest)"
else:
y_axis.label.val = "Orthogonal threshold (contrast units)"
elif cond == "lead":
if form == "rank":
y_axis.label.val = "Leading context effect (rank; 1 = lowest)"
else:
y_axis.label.val = "Leading context effect (contrast units)"
if form == "linear":
if cond == "sim":
y_max = 0.5
elif cond == "sim_orth":
y_max = 0.03
y_axis.TickLabels.format.val = "%.3g"
elif cond == "lead":
y_max = 0.085
y_axis.max.val = y_max
y_axis.min.val = 0.0
x_axis.min.val = -2
else:
x_axis.min.val = -10
x_axis.max.val = 105
y_axis.min.val = -10
y_axis.max.val = 105
x_axis.MajorTicks.manualTicks.val = [1] + range(20, 83, 20) + [93]
x_axis.MinorTicks.hide.val = True
y_axis.MajorTicks.manualTicks.val = [1] + range(20, 83, 20) + [93]
y_axis.MinorTicks.hide.val = True
if save_pdf:
_save(embed, conf, "ss_timing_{c:s}_{f:s}_scatter".format(c=cond,
f=form))
embed.EnableToolbar(True)
embed.WaitForClose()
def context_by_booth(save_pdf=False):
conf = ss_timing_analysis.conf.get_conf()
dem = ss_timing_analysis.dem.demographics()
booths = np.array(
[dem[subj_id]["testing_booth"] for subj_id in conf.subj_ids])
potential_booths = np.unique(booths)
n_booths = len(potential_booths)
assert n_booths == 2
# load the fit parameters, excluding the bad subjects
# this will be subj x onsets x oris x (a,b) x (est, 2.5, 97.5)
(fit, _, _) = ss_timing_analysis.group_fit.load_fit_data(exclude=True)
# restrict to just the alpha estimates
data = fit[..., 0, 0]
# and look at the context effect for simultaneous
data = data[:, 1, 1] - data[:, 1, 0]
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
page = embed.Root.Add("page")
page.width.val = "12cm"
page.height.val = "8cm"
graph = page.Add("graph", autoadd=False)
graph.bottomMargin.val = "1cm"
graph.topMargin.val = "0.6cm"
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
for i_booth in xrange(n_booths):
curr_booth = data[booths == potential_booths[i_booth]]
boxplot = graph.Add("boxplot")
dataset_str = "data_{b:d}".format(b=i_booth)
embed.SetData(dataset_str, curr_booth)
boxplot.values.val = dataset_str
boxplot.posn.val = i_booth
boxplot.labels.val = "{b:d} (n={n:d})".format(b=i_booth + 1,
n=len(curr_booth))
boxplot.fillfraction.val = 0.3
boxplot.markerSize.val = "2pt"
x_axis.mode.val = "labels"
x_axis.MajorTicks.manualTicks.val = [0, 1]
x_axis.MinorTicks.hide.val = True
x_axis.label.val = "Testing booth"
y_axis.TickLabels.format.val = "%.3g"
y_axis.min.val = 0.0
y_axis.max.val = 0.5
y_axis.label.val = "Context effect for simultaneous (par - orth)"
if save_pdf:
_save(embed, conf, "ss_timing_context_by_booth")
embed.EnableToolbar(True)
embed.WaitForClose()
def pairwise_corr(save_pdf=False):
conf = ss_timing_analysis.conf.get_conf()
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
page = embed.Root.Add("page")
page.width.val = "18cm"
page.height.val = "15cm"
grid = page.Add("grid")
grid.rows.val = 4
grid.columns.val = 4
grid.leftMargin.val = grid.rightMargin.val = "0cm"
grid.topMargin.val = "0cm"
grid.bottomMargin.val = "0.1cm"
# y_max = [20, 22, 15, 17]
ss_nice = [
"Unusual\\\\experiences", "Cognitive\\\\disorganisation",
"Introvertive\\\\anhedonia", "Impulsive\\\\nonconformity"
]
for (i_row, row_sub) in enumerate(
("un_ex", "cog_dis", "int_anh", "imp_non")):
row_ss = ss_timing_analysis.dem.get_olife_subscale(row_sub,
exclude=True)
row_ss = scipy.stats.rankdata(row_ss)
for (i_col, col_sub) in enumerate(
("un_ex", "cog_dis", "int_anh", "imp_non")):
col_ss = ss_timing_analysis.dem.get_olife_subscale(col_sub,
exclude=True)
col_ss = scipy.stats.rankdata(col_ss)
graph = grid.Add("graph", autoadd=False)
graph.leftMargin.val = graph.rightMargin.val = "0.3cm"
graph.bottomMargin.val = graph.topMargin.val = "0.3cm"
graph.aspect.val = 1
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
if i_col < i_row:
print scipy.stats.pearsonr(row_ss, col_ss)
xy = graph.Add("xy")
xy.xData.val = row_ss
xy.yData.val = col_ss
xy.PlotLine.hide.val = True
xy.MarkerFill.transparency.val = 60
xy.MarkerLine.hide.val = True
xy.markerSize.val = "2pt"
# x_axis.label.val = ss_nice[i_row]
# y_axis.label.val = ss_nice[i_col]
x_axis.min.val = -10
x_axis.max.val = 105
y_axis.min.val = -10
y_axis.max.val = 105
x_axis.MajorTicks.manualTicks.val = [1] + range(20, 83,
20) + [93]
x_axis.MinorTicks.hide.val = True
y_axis.MajorTicks.manualTicks.val = [1] + range(20, 83,
20) + [93]
y_axis.MinorTicks.hide.val = True
else:
for ax in (x_axis, y_axis):
ax.autoMirror.val = True
ax.MajorTicks.hide.val = True
ax.MinorTicks.hide.val = True
ax.TickLabels.hide.val = True
label = graph.Add("label")
label.alignHorz.val = "centre"
label.alignVert.val = "centre"
if i_row == i_col:
label.label.val = ss_nice[i_row]
else:
(r, p) = scipy.stats.pearsonr(row_ss, col_ss)
r = np.round(r, 2)
r_str = "{r:.02f}".format(r=r)
p = np.round(p, 3)
if p < 0.001:
p_str = "< 0.001"
else:
p_str = "= {p:.03f}".format(p=p)
label.label.val = "\\textit{{r}} = {r:s}\\\\ \\textit{{p}} {p:s}".format(
r=r_str, p=p_str)
if save_pdf:
_save(embed, conf, "ss_timing_pairwise")
embed.EnableToolbar(True)
embed.WaitForClose()
def norms_comparison(save_pdf=False):
norms = {
"un_ex": {
"F": {
0.25: 4,
0.5: 9,
0.75: 15,
0.9: 19.2
},
"M": {
0.25: 4,
0.5: 9,
0.75: 15,
0.9: 19
}
},
"cog_dis": {
"F": {
0.25: 8,
0.5: 13,
0.75: 17,
0.9: 21
},
"M": {
0.25: 8,
0.5: 12,
0.75: 16,
0.9: 20
}
},
"int_anh": {
"F": {
0.25: 2,
0.5: 4,
0.75: 7,
0.9: 10
},
"M": {
0.25: 2,
0.5: 5,
0.75: 8,
0.9: 11
}
},
"imp_non": {
"F": {
0.25: 6,
0.5: 9,
0.75: 12,
0.9: 14
},
"M": {
0.25: 6,
0.5: 10,
0.75: 13,
0.9: 15
}
}
}
conf = ss_timing_analysis.conf.get_conf()
dem = ss_timing_analysis.dem.demographics()
genders = np.array([dem[subj_id]["gender"] for subj_id in conf.subj_ids])
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
page = embed.Root.Add("page")
page.width.val = "18cm"
page.height.val = "15cm"
grid = page.Add("grid")
grid.rows.val = 2
grid.columns.val = 2
grid.leftMargin.val = grid.rightMargin.val = "0cm"
grid.topMargin.val = "0cm"
grid.bottomMargin.val = "0.1cm"
y_max = [20, 22, 15, 17]
ss_nice = [
"Unusual experiences", "Cognitive disorganisation",
"Introvertive anhedonia", "Impulsive nonconformity"
]
for (i_sub, subscale) in enumerate(
("un_ex", "cog_dis", "int_anh", "imp_non")):
curr_ss = ss_timing_analysis.dem.get_olife_subscale(subscale,
exclude=True)
graph = grid.Add("graph", autoadd=False)
graph.bottomMargin.val = "1cm"
graph.topMargin.val = "0.6cm"
label = graph.Add("label")
label.label.val = ss_nice[i_sub]
label.yPos.val = 1.01
label.xPos.val = 0.5
label.alignHorz.val = "centre"
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
for (i_gender, gender) in enumerate(["F", "M"]):
# first, the norms
boxplot = graph.Add("boxplot")
boxplot.calculate.val = False
boxplot.median.val = norms[subscale][gender][0.5]
boxplot.boxmin.val = norms[subscale][gender][0.25]
boxplot.boxmax.val = norms[subscale][gender][0.75]
boxplot.whiskermax.val = norms[subscale][gender][0.9]
boxplot.whiskermin.val = norms[subscale][gender][0.25]
boxplot.mean.val = norms[subscale][gender][0.5]
boxplot.posn.val = i_gender - 0.15
boxplot.fillfraction.val = 0.15
boxplot.markerSize.val = "2pt"
boxplot.labels.val = "Norm\\\\({g:s})".format(g=gender)
boxplot.meanmarker.val = "none"
curr_gender = curr_ss[genders == gender]
(b25, b50, b75,
b90) = scipy.stats.scoreatpercentile(curr_gender,
[25, 50, 75, 90])
boxplot = graph.Add("boxplot")
boxplot.calculate.val = False
boxplot.median.val = b50
boxplot.boxmin.val = b25
boxplot.boxmax.val = b75
boxplot.whiskermax.val = b90
boxplot.whiskermin.val = b25
boxplot.mean.val = b50
boxplot.posn.val = i_gender + 0.15
boxplot.fillfraction.val = 0.15
boxplot.markerSize.val = "2pt"
boxplot.labels.val = "Curr\\\\({g:s})".format(g=gender)
boxplot.meanmarker.val = "none"
y_axis.min.val = 0.0
y_axis.max.val = y_max[i_sub]
y_axis.MajorTicks.manualTicks.val = range(0, y_max[i_sub] + 1, 5)
x_axis.mode.val = "labels"
x_axis.MajorTicks.manualTicks.val = [-0.15, 0.15, 0.85, 1.15]
x_axis.MinorTicks.hide.val = True
x_axis.label.val = "Source and gender"
if save_pdf:
_save(embed, conf, "ss_timing_norms_comparison")
embed.EnableToolbar(True)
embed.WaitForClose()
def scatter_sub(save_pdf=False):
conf = ss_timing_analysis.conf.get_conf()
# load the fit parameters, excluding the bad subjects
# this will be subj x onsets x oris x (a,b) x (est, 2.5, 97.5)
(fit, _, _) = ss_timing_analysis.group_fit.load_fit_data(exclude=True)
# restrict to just the alpha estimates
data = fit[..., 0, 0]
data = data[:, 1, 1] - data[:, 1, 0]
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
page = embed.Root.Add("page")
page.width.val = "18cm"
page.height.val = "15cm"
grid = page.Add("grid")
grid.rows.val = 2
grid.columns.val = 2
grid.leftMargin.val = grid.rightMargin.val = "0cm"
grid.topMargin.val = "0cm"
grid.bottomMargin.val = "0.1cm"
ss_nice = [
"Unusual experiences", "Cognitive disorganisation",
"Introvertive anhedonia", "Impulsive nonconformity"
]
for (i_sub, subscale) in enumerate(
("un_ex", "cog_dis", "int_anh", "imp_non")):
curr_ss = ss_timing_analysis.dem.get_olife_subscale(subscale,
exclude=True)
assert len(curr_ss) == len(data)
graph = grid.Add("graph", autoadd=False)
graph.bottomMargin.val = "1cm"
graph.topMargin.val = "0.6cm"
label = graph.Add("label")
label.label.val = ss_nice[i_sub]
label.yPos.val = 1.01
label.xPos.val = 0.5
label.alignHorz.val = "centre"
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
xy = graph.Add("xy")
xy.xData.val = curr_ss
xy.yData.val = data
xy.PlotLine.hide.val = True
xy.MarkerFill.transparency.val = 60
xy.MarkerLine.hide.val = True
x_axis.label.val = "Score"
y_axis.label.val = "Context effect for simultaneous (par - orth)"
y_max = 0.5
y_axis.max.val = y_max
y_axis.min.val = 0.0
x_axis.min.val = -2
if save_pdf:
_save(embed, conf, "ss_timing_scatter_sub")
embed.EnableToolbar(True)
embed.WaitForClose()
def old_scatter(save_pdf=False, cond="sim"):
conf = ss_timing_analysis.conf.get_conf()
# load the fit parameters, excluding the bad subjects
# this will be subj x onsets x oris x (a,b) x (est, 2.5, 97.5)
(fit, _, _) = ss_timing_analysis.group_fit.load_fit_data(exclude=True)
# restrict to just the alpha estimates
data = fit[..., 0, 0]
if cond == "sim":
data = data[:, 1, 1] - data[:, 1, 0]
elif cond == "sim_orth":
data = data[:, 1, 0]
elif cond == "lead":
data = data[:, 0, 1] - data[:, 0, 0]
dem = ss_timing_analysis.dem.demographics()
# righto, now for the sz scores
sz = np.array([dem[subj_id]["olife_total"] for subj_id in conf.subj_ids])
# check that we've exlcuded subjects, as we think we should have
assert len(sz) == len(data)
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
page = embed.Root.Add("page")
page.width.val = "18cm"
page.height.val = "8cm"
grid = page.Add("grid")
grid.rows.val = 1
grid.columns.val = 2
grid.leftMargin.val = grid.rightMargin.val = "0cm"
grid.bottomMargin.val = grid.topMargin.val = "0cm"
grid.scaleCols.val = [0.85, 0.17]
# SCATTER
graph = grid.Add("graph", autoadd=False)
graph.bottomMargin.val = "1cm"
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
xy = graph.Add("xy")
xy.xData.val = sz
xy.yData.val = data
xy.PlotLine.hide.val = True
xy.MarkerFill.transparency.val = 60
xy.MarkerLine.hide.val = True
x_axis.label.val = "Schizotypy score"
if cond == "sim":
y_axis.label.val = "Context effect for simultaneous (par - orth)"
y_max = 0.5
elif cond == "sim_orth":
y_axis.label.val = "Contrast detection threshold for simultaneous, orth"
y_max = 0.03
y_axis.TickLabels.format.val = "%.3g"
elif cond == "lead":
y_axis.label.val = "Context effect for leading surround (par - orth)"
y_max = 0.085
y_axis.max.val = y_max
y_axis.min.val = 0.0
# KDE
graph = grid.Add("graph", autoadd=False)
graph.leftMargin.val = "0cm"
graph.bottomMargin.val = "1cm"
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
kde = scipy.stats.gaussian_kde(data)
kde_x = np.linspace(0, y_max, 100)
kde_y = kde(kde_x)
xy = graph.Add("xy")
xy.xData.val = kde_y
xy.yData.val = kde_x
xy.MarkerFill.hide.val = True
xy.MarkerLine.hide.val = True
xy.FillBelow.fillto.val = "left"
xy.FillBelow.color.val = "grey"
xy.PlotLine.color.val = "grey"
xy.FillBelow.hide.val = False
y_axis.max.val = y_max
x_axis.hide.val = True
x_axis.lowerPosition.val = 0.075
if save_pdf:
_save(embed, conf, "ss_timing_{c:s}_scatter".format(c=cond))
embed.EnableToolbar(True)
embed.WaitForClose()
def split(save_pdf=False):
conf = ss_timing_analysis.conf.get_conf()
split_locs = [50, 40, 30, 20, 10]
n_splits = len(split_locs)
data = ss_timing_analysis.split.split_ana()
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
page = embed.Root.Add("page")
page.width.val = "10cm"
page.height.val = "10cm"
graph = page.Add("graph", autoadd=False)
graph.bottomMargin.val = "2.5cm"
graph.topMargin.val = "0.2cm"
graph.leftMargin.val = "1.5cm"
graph.rightMargin.val = "0cm"
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
subscale_colours = ["#e41a1c", "#377eb8", "#4daf4a", "#984ea3"]
split_lbl = [
"< {s1:d}%, >= {s2:d}%".format(s1=sloc, s2=100 - sloc)
for sloc in split_locs
]
embed.SetDataText("split_lbl", split_lbl)
for (i_sub, sub) in enumerate(conf.subscales):
embed.SetData(sub, data[:, i_sub, 0], symerr=data[:, i_sub, 1])
xy = graph.Add("xy")
x = np.arange(n_splits) + np.linspace(-0.15, 0.15, 4)[i_sub]
xy.xData.val = x
xy.yData.val = sub
if i_sub == 0:
xy.labels.val = "split_lbl"
xy.MarkerLine.hide.val = False
#xy.MarkerLine.transparency.val = 20
xy.MarkerFill.hide.val = True
#xy.PlotLine.hide.val = True
xy.markerSize.val = "2pt"
xy.MarkerLine.color.val = subscale_colours[i_sub]
xy.PlotLine.color.val = subscale_colours[i_sub]
xy.ErrorBarLine.color.val = subscale_colours[i_sub]
#xy.marker.val = ana_markers[i_ana]
xy.key.val = "".join(sub.title().split("_"))
xy.Label.hide.val = True
key = graph.Add("key")
key.horzPosn.val = "manual"
key.vertPosn.val = "manual"
key.horzManual.val = 0.051125165504591
key.vertManual.val = 0.79086156527087
x_axis.mode.val = "labels"
x_axis.min.val = -0.5
x_axis.max.val = 4.5
x_axis.MinorTicks.hide.val = True
x_axis.MajorTicks.manualTicks.val = range(n_splits)
x_axis.TickLabels.rotate.val = "45"
x_axis.label.val = "Group split"
y_axis.label.val = "Context effect difference (high - low)"
if save_pdf:
_save(embed, conf, "ss_timing_split")
embed.EnableToolbar(True)
embed.WaitForClose()
def bf(save_pdf=False):
conf = ss_timing_analysis.conf.get_conf()
conf.n_boot = 1
# N x (r, bf)
r_bf = np.loadtxt(os.path.join(conf.base_path, "ss_timing_bf.txt"))
# analysis (3) x subscale (4)
corr = ss_timing_analysis.stats.correlations(conf)[..., 0]
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
page = embed.Root.Add("page")
page.width.val = "8cm"
page.height.val = "8.5cm"
graph = page.Add("graph", autoadd=False)
graph.bottomMargin.val = "2.5cm"
graph.topMargin.val = "0.2cm"
graph.leftMargin.val = "0.9cm"
graph.rightMargin.val = "1.4cm"
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
subscale_colours = ["#e41a1c", "#377eb8", "#4daf4a", "#984ea3"]
ana_markers = ["circle", "square", "triangle"]
ana_sizes = [4, 3.54, 5.01]
for i_sub in xrange(4):
for i_ana in xrange(3):
xy = graph.Add("xy")
xy.xData.val = corr[i_ana, i_sub]
# need to find the nearest bayes factor
i_bf = np.argmin(np.abs(corr[i_ana, i_sub] - r_bf[:, 0]))
bf = r_bf[i_bf, 1]
print i_sub, i_ana, np.log(bf)
xy.yData.val = np.log(bf)
xy.MarkerLine.hide.val = False
#xy.MarkerLine.transparency.val = 20
xy.MarkerFill.hide.val = True
xy.PlotLine.hide.val = True
xy.markerSize.val = "{s:.3f}pt".format(s=ana_sizes[i_ana] * 0.7)
xy.MarkerLine.color.val = subscale_colours[i_sub]
xy.marker.val = ana_markers[i_ana]
xy.key.val = " "
bf_xy = graph.Add("xy")
bf_xy.xData.val = r_bf[:, 0]
bf_xy.yData.val = np.log(r_bf[:, 1])
bf_xy.MarkerFill.hide.val = True
bf_xy.MarkerLine.hide.val = True
log_div = np.log([30.0, 10.0, 3.0, 1.0, 1 / 3.0, 1 / 10.0, 1 / 30.])
div_labels = [
"Strong H_1", "Substantial H_1", "Anecdotal H_1", "Anecdotal H_0",
"Substantial H_0", "Strong H_0", ""
]
div_delta = (log_div[0] - log_div[1]) / 2.0
for (curr_div, div_lbl) in zip(log_div, div_labels):
xy = graph.Add("xy")
xy.xData.val = [-0.4, 0.4]
xy.yData.val = [curr_div] * 2
xy.MarkerFill.hide.val = True
xy.MarkerLine.hide.val = True
xy.PlotLine.style.val = "dash1"
xy.color.val = "grey"
if div_lbl != "":
lbl = graph.Add("label")
lbl.label.val = div_lbl
lbl.xPos.val = 0.361
lbl.yPos.val = curr_div - div_delta
lbl.alignVert.val = "centre"
lbl.positioning.val = "axes"
hoff = 0.1
key = graph.Add("key")
key.Border.hide.val = True
key.keyLength.val = "0.415cm"
key.columns.val = 4
key.horzPosn.val = "manual"
key.vertPosn.val = "manual"
key.horzManual.val = 0.188 + hoff
key.vertManual.val = -0.397
col_start = 0.224 + hoff
col_delta = 0.409 - 0.224
for (i_col, ss) in enumerate(conf.subscales):
lbl = graph.Add("label")
lbl.label.val = "".join(ss.title().split("_"))
lbl.xPos.val = col_start + i_col * col_delta
lbl.yPos.val = -0.233
lbl.alignHorz.val = "centre"
lbl.Text.size.val = "7pt"
ana_labels = ["Context (sim.)", "Orthogonal (sim.)", "Context (delay)"]
row_start = -0.293
row_delta = -0.293 - -0.341
for (i_row, ana_label) in enumerate(ana_labels):
lbl = graph.Add("label")
lbl.label.val = ana_label
lbl.xPos.val = 0.151 + hoff
lbl.yPos.val = row_start - i_row * row_delta
lbl.alignHorz.val = "right"
lbl.Text.size.val = "7pt"
x_axis.label.val = "Correlation (r)"
x_axis.TickLabels.format.val = "%.1f"
x_axis.Label.offset.val = "2pt"
y_axis.min.val = -4
y_axis.max.val = 4
y_axis.label.val = "log(Bayes Factor)"
y_axis.TickLabels.format.val = "%.1f"
y_axis.Label.offset.val = "2pt"
y_axis.MinorTicks.hide.val = True
if save_pdf:
_save(embed, conf, "ss_timing_bf")
embed.EnableToolbar(True)
embed.WaitForClose()
def subjects(save_pdf=False):
conf = ss_timing_analysis.conf.get_conf()
# subj x onsets x oris x bins x (prop, n)
data = ss_timing_analysis.group_data.bin_group_data()
# fit is: subj x onsets x oris x (a, b) x (est, 2.5, 97.5)
# fit_fine is: subj x onsets x oris x X x 2 (2.5, 97.5)
(fit, fit_fine, _) = ss_timing_analysis.group_fit.load_fit_data()
embed = veusz.embed.Embedded("veusz")
figutils.set_veusz_style(embed)
embed.SetData("bin_centres", conf.bin_centres)
embed.SetData("fine_x", conf.fine_x)
for (i_subj, subj_id) in enumerate(conf.all_subj_ids):
page = embed.Root.Add("page")
page.width.val = "21cm"
page.height.val = "29.7cm"
label = page.Add("label")
label.label.val = subj_id
label.yPos.val = 0.77
if subj_id in conf.exclude_ids:
label.label.val += " (excluded)"
label.Text.color.val = "red"
grid = page.Add("grid")
grid.rows.val = 2
grid.columns.val = 2
grid.leftMargin.val = grid.rightMargin.val = "3cm"
grid.topMargin.val = "8cm"
grid.bottomMargin.val = "8cm"
for i_onset in xrange(conf.n_surr_onsets):
for i_ori in xrange(conf.n_surr_oris):
graph = grid.Add("graph", autoadd=False)
x_axis = graph.Add("axis")
y_axis = graph.Add("axis")
cond_label = graph.Add("label")
cond_label.label.val = ", ".join([
conf.surr_ori_labels[i_ori],
conf.surr_onset_labels[i_onset]
])
cond_label.yPos.val = 1.02
cond_label.xPos.val = 0.5
cond_label.alignHorz.val = "centre"
cond_label.Text.size.val = "8pt"
# CROSSHAIRS
pse_y = graph.Add("xy")
pse_y.xData.val = [0.001, fit[i_subj, i_onset, i_ori, 0, 0]]
pse_y.yData.val = [
conf.psych_func(fit[i_subj, i_onset, i_ori, 0,
0], fit[i_subj, i_onset, i_ori, 0, 0],
fit[i_subj, i_onset, i_ori, 0, 1])
] * 2
pse_x = graph.Add("xy")
pse_x.xData.val = [fit[i_subj, i_onset, i_ori, 0, 0]] * 2
pse_x.yData.val = [
-0.05,
conf.psych_func(fit[i_subj, i_onset, i_ori, 0,
0], fit[i_subj, i_onset, i_ori, 0, 0],
fit[i_subj, i_onset, i_ori, 0, 1])
]
for pse_ax in (pse_y, pse_x):
pse_ax.MarkerFill.hide.val = True
pse_ax.MarkerLine.hide.val = True
pse_ax.PlotLine.style.val = "dashed"
# POINTS
points = graph.Add("xy")
prop_name = "resp_prop_{s:d}_{t:d}_{o:d}".format(s=i_subj,
t=i_onset,
o=i_ori)
embed.SetData(prop_name, data[i_subj, i_onset, i_ori, :, 0])
k_name = "resp_k_{s:d}_{t:d}_{o:d}".format(s=i_subj,
t=i_onset,
o=i_ori)
point_scale = np.sqrt(
data[i_subj, i_onset, i_ori, :, 1] / np.pi) * 2 * 0.35
embed.SetData(k_name, point_scale)
points.xData.val = "bin_centres"
points.yData.val = prop_name
points.scalePoints.val = k_name
points.MarkerLine.hide.val = True
points.MarkerFill.transparency.val = 50
points.PlotLine.hide.val = True
points.MarkerFill.color.val = "blue"
# FIT
fit_plot = graph.Add("xy")
fit_name = "fit_{s:d}_{t:d}_{o:d}".format(s=i_subj,
t=i_onset,
o=i_ori)
fit_y = conf.psych_func(conf.fine_x,
alpha=fit[i_subj, i_onset, i_ori, 0,
0],
beta=fit[i_subj, i_onset, i_ori, 1, 0])
embed.SetData(
fit_name,
fit_y,
poserr=np.abs(fit_fine[i_subj, i_onset, i_ori, :, 1] -
fit_y),
negerr=np.abs(fit_fine[i_subj, i_onset, i_ori, :, 0] -
fit_y))
fit_plot.xData.val = "fine_x"
fit_plot.yData.val = fit_name
fit_plot.MarkerFill.hide.val = True
fit_plot.MarkerLine.hide.val = True
fit_plot.errorStyle.val = "fillvert"
fit_plot.ErrorBarLine.hide.val = True
x_axis.log.val = True
x_axis.label.val = "Contrast"
x_axis.TickLabels.format.val = "%.3g"
x_axis.MajorTicks.manualTicks.val = [0.001, 0.01, 0.1, 0.5, 1]
y_axis.min.val = -0.1
y_axis.max.val = 1.1
y_axis.label.val = "Accuracy (prop. correct)"
y_axis.MinorTicks.hide.val = True
y_axis.MajorTicks.manualTicks.val = [0, 0.25, 0.5, 0.69, 1]
y_axis.TickLabels.format.val = "%.02f"
if save_pdf:
_save(embed, conf, "ss_timing_subjects", page=range(conf.n_all_subj))
embed.EnableToolbar(True)
embed.WaitForClose()
