def plot_pooled_correlation_dists_by_condition(cfg):
"""
Plot the pooled (ie. combined, pooled) correlation distributions
separately for each of the conditions (each list in fname_group_list)
corresponds to a condition.
Parameters
----------
cfg : dict
the user-specified config dictionary.
The following keys are required::
"group_1_mat_fnames"
"group_2_mat_fnames"
"group_1_color"
"group_2_color"
"group_1_label"
"group_2_label"
Returns
-------
fig : the matplotlib figure object
"""
# CFGCHANGE?
config.require(cfg, [
"group_1_mat_fnames", "group_2_mat_fnames", "group_1_color",
"group_2_color", "group_1_label", "group_2_label"
])
fname_group_list = [cfg["group_1_mat_fnames"], cfg["group_2_mat_fnames"]]
colors = [cfg["group_1_color"], cfg["group_2_color"]]
labels = [cfg["group_1_label"], cfg["group_2_label"]]
n_bins = 100
corr_bins, corr_bin_centers = aux.get_lin_bins(n_bins, -1, 1.)
bin_counts = [np.zeros(n_bins) for f in fname_group_list]
fig = plt.figure()
ax = fig.add_subplot(111)
for i, fname_group in enumerate(fname_group_list):
for fname in fname_group:
flat_corr_mat = \
dataio.get_blacklist_filtered_and_flattened_adj_mat(
fname, cfg["blacklist_fname"]
)
bin_counts[i] += aux.get_bin_counts(flat_corr_mat, corr_bins)
# normalize
bin_counts[i] = bin_counts[i] / \
(np.sum(bin_counts[i] * (corr_bins[1] - corr_bins[0])))
ax.plot(corr_bin_centers,
bin_counts[i],
color=colors[i],
label=labels[i])
ax.set_xlabel(settings.get_prop_tex_name(settings.correlation_tag))
ax.set_ylabel(r"Probability density P(c)")
ax.legend(loc=0)
fig.savefig(cfg['outdata_dir'] + "pooledCorrDists.pdf",
format="pdf",
bbox_inches='tight')
return fig
def plot_pooled_corr_t_val_dists(cfg):
"""
Plot the tvalue distributions for movie and rest
Parameters
----------
config : dict
The following keys are required::
"group_1_mat_fnames"
"group_2_mat_fnames"
"group_1_color"
"group_2_color"
"group_1_label"
"group_2_label"
"outdata_dir"
"paired"
Returns
-------
fig : the matplotlib figure object
"""
config.require(cfg, [
"group_1_mat_fnames", "group_2_mat_fnames", "group_1_color",
"group_2_color", "group_1_label", "group_2_label", "outdata_dir",
"paired"
])
# get tvals
flat_mats = dataio.get_blacklist_filtered_and_flattened_adj_mats(
cfg["all_fnames"], cfg["blacklist_fname"])
if cfg["paired"]:
t_vals = measures.paired_t_value(flat_mats,
len(cfg['group_1_mat_fnames']))
else:
t_vals = measures.unpaired_t_value(flat_mats,
len(cfg['group_1_mat_fnames']))
minVal = np.min(t_vals)
maxVal = np.max(t_vals)
n_bins = 100
bins, binCenters = aux.get_lin_bins(n_bins,
minVal - (np.abs(minVal) * 0.1),
maxVal + (np.abs(maxVal) * 0.1))
bin_counts = aux.get_bin_counts(t_vals, bins)
fig = plt.figure()
ax = fig.add_subplot(111)
# normalize
bin_counts = bin_counts * 1. / (np.sum(bin_counts) * (bins[1] - bins[0]))
labels = cfg['group_1_label'], cfg["group_2_label"]
ax.plot(binCenters, bin_counts, label=r"" + labels[0] + "-" + labels[1])
ax.set_xlabel(settings.get_prop_tex_name(settings.tval_tag))
ax.set_ylabel(r"Probability density")
ax.legend(loc=0)
plt.savefig(cfg["outdata_dir"] + "tvalDist.pdf",
format="pdf",
bbox_inches='tight')
return fig
def plot_corr_dists_subjectwise(cfg):
"""
Plotting the individual correlation profiles to see the variations.
Parameters
----------
cfg : dict
the user-specified config dictionary.
The following keys are required::
"group_1_mat_fnames"
"group_2_mat_fnames"
"group_1_color"
"group_2_color"
"group_1_label"
"group_2_label"
"outdata_dir"
Returns
-------
fig : the matplotlib figure object
"""
fname_group_list = [cfg["group_1_mat_fnames"], cfg["group_2_mat_fnames"]]
colors = [cfg["group_1_color"], cfg["group_2_color"]]
labels = [cfg["group_1_label"], cfg["group_2_label"]]
rc('legend', fontsize=8)
fig = plt.figure()
n_bins = 100
n = len(fname_group_list[0])
ax_x, ax_y = _get_subplot_x_y(n)
for i, _ in enumerate(fname_group_list[0]): # now n1 is n2)
ax = fig.add_subplot(ax_y, ax_x, i + 1)
for j, fname_group in enumerate(fname_group_list):
corr_bins, corr_bin_centers = aux.get_lin_bins(n_bins, -1, 1.)
flat_corr_mat = \
dataio.get_blacklist_filtered_and_flattened_adj_mat(
fname_group[i], cfg["blacklist_fname"]
)
bin_counts = aux.get_bin_counts(flat_corr_mat, corr_bins)
# normalize
bin_counts = bin_counts / \
(np.sum(bin_counts * (corr_bins[1] - corr_bins[0])))
ax.plot(corr_bin_centers,
bin_counts,
color=colors[j],
label=labels[j] + "\_" + str(i))
# ax.set_xlabel(settings.get_prop_tex_name("corr"))
# ax.set_ylabel(r"Probability density P(c)")
ax.legend(loc=0)
plt.tight_layout()
fig.savefig(cfg["outdata_dir"] + "individualCorrDistsSubjectwise.pdf",
format="pdf",
bbox_inches='tight')
return fig
def plot_individual_correlation_dists(cfg):
"""
Plotting the individual correlation profiles to see the variations.
Parameters
----------
cfg : dict
the user-specified config dictionary.
The following keys are required::
"group_1_mat_fnames"
"group_2_mat_fnames"
"group_1_color"
"group_2_color"
"group_1_label"
"group_2_label"
"outdata_dir"
Returns
-------
fig : the matplotlib figure object
"""
# CFGCHANGE?
config.require(cfg, )
fname_group_list = [cfg["group_1_mat_fnames"], cfg["group_2_mat_fnames"]]
colors = [cfg["group_1_color"], cfg["group_2_color"]]
labels = [cfg["group_1_label"], cfg["group_2_label"]]
fig = plt.figure()
ax = fig.add_subplot(111)
n_bins = 100
corr_bins, corr_bin_centers = aux.get_lin_bins(n_bins, -1, 1.)
for i, fname_group in enumerate(fname_group_list):
for fname in fname_group:
flat_corr_mat = \
dataio.get_blacklist_filtered_and_flattened_adj_mat(
fname, cfg['blacklist_fname']
)
bin_counts = aux.get_bin_counts(flat_corr_mat, corr_bins)
# normalize
bin_counts = bin_counts / \
(np.sum(bin_counts * (corr_bins[1] - corr_bins[0])))
ax.plot(corr_bin_centers,
bin_counts,
color=colors[i],
label=labels[i])
ax.set_xlabel(settings.get_prop_tex_name("corr"))
ax.set_ylabel(r"Probability density P(c)")
ax.legend(loc=0)
fig.savefig(cfg['outdata_dir'] + "individualCorrDists.pdf",
format="pdf",
bbox_inches='tight')
return fig
def plot_lin_pdf(ax,
values,
n_bins=100,
color="r",
label=None,
normalize=True):
"""
Plots a PDF with linear bins.
Parameters
----------
ax : matplotlib axes object
the axes to plot to
values : 1D numpy array
the values of the pdf
n_bins : int, optional
the number of bins in the pdf
color : any valid matplotlib color, optional
the color which the pdf is plotted with
label : str, optional
the label of the pdf
normalize : bool, optional
if False, the counts are shown instead of the pdf
"""
values = np.array(values)
val_max = np.max(values)
val_min = np.min(values)
linbins, bin_centers = aux.get_lin_bins(n_bins,
val_min - 0.1 * np.abs(val_min),
val_max + 0.1 * np.abs(val_max))
bin_counts = np.bincount(np.digitize(values, linbins),
minlength=n_bins + 1)[1:]
# normalize
if normalize:
bin_counts = bin_counts * 1. / \
(len(values) * (linbins[1] - linbins[0]))
if label is not None:
ax.plot(bin_centers, bin_counts, color=color, label=label)
else:
ax.plot(bin_centers, bin_counts, color=color)