def visualize_importance(importance, ax, xticklabels=True,
yticklabels=True, pad=0, ymax=None, legend=True):
"""Plot task loadings on one axis"""
importance_vars = importance[0]
importance_vals = [abs(i)+pad for i in importance[1].T]
plot_loadings(ax, importance_vals, kind='line', offset=.5,
plot_kws={'alpha': 1, 'linewidth': 3})
# set up x ticks
xtick_locs = np.arange(0.0, 2*np.pi, 2*np.pi/len(importance_vars))
ax.set_xticks(xtick_locs)
ax.set_xticks(xtick_locs+np.pi/len(importance_vars), minor=True)
if xticklabels:
if type(importance_vars[0]) == str:
ax.set_xticklabels(importance_vars,
y=.08, minor=True)
else:
ax.set_xticklabels(['Fac %s' % str(i+1) for i in importance_vars],
y=.08, minor=True)
# set up yticks
if ymax:
ax.set_ylim(top=ymax)
ytick_locs = ax.yaxis.get_ticklocs()
new_yticks = np.linspace(0, ytick_locs[-1], 7)
ax.set_yticks(new_yticks)
if yticklabels:
labels = np.round(new_yticks,2)
replace_dict = {i:'' for i in labels[::2]}
labels = [replace_dict.get(i, i) for i in labels]
ax.set_yticklabels(labels)
if legend:
ax.legend(loc='upper center', bbox_to_anchor=(.5,-.15))
def visualize_importance(importance, ax, xticklabels=True, yticklabels=True,
axes_linewidth=None, label_size=10,
label_scale=0, title=None,
ymax=None, color='k'):
importance_vars = importance[0]
importance_vars = [shortened_factors.get(v,v) for v in importance_vars]
if importance[1] is not None:
importance_vals = [abs(i) for i in importance[1]]
plot_loadings(ax, importance_vals, kind='line', offset=.5,
colors=[color], plot_kws={'alpha': 1, 'linewidth': label_size/4})
else:
ax.set_yticks([])
ax.grid(linewidth=label_size/8)
if axes_linewidth:
plt.setp(list(ax.spines.values()), linewidth=axes_linewidth)
# set up x ticks
xtick_locs = np.arange(0.0, 2*np.pi, 2*np.pi/len(importance_vars))
ax.set_xticks(xtick_locs)
ax.set_xticks(xtick_locs+np.pi/len(importance_vars), minor=True)
if xticklabels:
if type(importance_vars[0]) != str:
importance_vars = [str(i+1) for i in importance_vars]
scale = 1+label_scale
size = ax.get_position().expanded(scale, scale)
ax2=ax.get_figure().add_axes(size,zorder=2)
for i, var in enumerate(importance_vars):
arc_start = (i+.15)*2*np.pi/len(importance_vars)
arc_end = (i+.85)*2*np.pi/len(importance_vars)
curve = [
.85*np.cos(np.linspace(arc_start,arc_end,100)),
.85*np.sin(np.linspace(arc_start,arc_end,100))
]
plt.plot(*curve, alpha=0)
curvetext = CurvedText(
x = curve[0][::-1],
y = curve[1][::-1],
text=var, #'this this is a very, very long text',
va = 'bottom',
axes = ax2,
fontsize=label_size
)
ax2.set_xlim([-1,1]); ax2.set_ylim([-1,1])
ax2.axis('off')
if title:
ax.set_title(title, fontsize=label_size*1.5, y=1.1)
# set up yticks
if len(importance[1]) != 0:
ax.set_ylim(bottom=0)
if ymax is None:
ymax = np.max(importance_vals)*1.1
ax.set_ylim(top=ymax)
new_yticks = np.linspace(0, ymax, 7)
ax.set_yticks(new_yticks)
if yticklabels:
labels = np.round(new_yticks,2)
replace_dict = {i:'' for i in labels[::2]}
labels = [replace_dict.get(i, i) for i in labels]
ax.set_yticklabels(labels)
def plot_cluster_factors(results,
c,
rotate='oblimin',
ext='png',
plot_dir=None):
"""
Args:
EFA: EFA_Analysis object
c: number of components for EFA
task_sublists: a dictionary whose values are sets of tasks, and
whose keywords are labels for those lists
"""
# set up variables
HCA = results.HCA
EFA = results.EFA
names, cluster_loadings = zip(
*HCA.get_cluster_loading(EFA, rotate=rotate).items())
cluster_DVs = HCA.get_cluster_DVs(inp='EFA%s_%s' % (EFA.get_c(), rotate))
cluster_loadings = list(
zip([cluster_DVs[n] for n in names], cluster_loadings))
max_loading = max([max(abs(i[1])) for i in cluster_loadings])
# plot
colors = sns.hls_palette(len(cluster_loadings))
ncols = min(5, len(cluster_loadings))
nrows = ceil(len(cluster_loadings) / ncols)
f, axes = plt.subplots(nrows,
ncols,
figsize=(ncols * 10, nrows * (8 + nrows)),
subplot_kw={'projection': 'polar'})
axes = f.get_axes()
for i, (measures, loading) in enumerate(cluster_loadings):
plot_loadings(axes[i],
loading,
kind='line',
offset=.5,
plot_kws={
'alpha': .8,
'c': colors[i]
})
axes[i].set_title('Cluster %s' % i, y=1.14, fontsize=25)
# set tick labels
xtick_locs = np.arange(0.0, 2 * np.pi, 2 * np.pi / len(loading))
axes[i].set_xticks(xtick_locs)
axes[i].set_xticks(xtick_locs + np.pi / len(loading), minor=True)
if i % (ncols * 2) == 0 or i % (ncols * 2) == (ncols - 1):
axes[i].set_xticklabels(loading.index, y=.08, minor=True)
# set ylim
axes[i].set_ylim(top=max_loading)
for j in range(i + 1, len(axes)):
axes[j].set_visible(False)
plt.subplots_adjust(hspace=.5, wspace=.5)
filename = 'polar_factors_EFA%s_%s.%s' % (c, rotate, ext)
if plot_dir is not None:
save_figure(f, path.join(plot_dir, filename), {'bbox_inches': 'tight'})
plt.close()
def plot_cluster_factors(results, c, rotate='oblimin', ext='png', plot_dir=None):
"""
Args:
EFA: EFA_Analysis object
c: number of components for EFA
task_sublists: a dictionary whose values are sets of tasks, and
whose keywords are labels for those lists
"""
# set up variables
HCA = results.HCA
EFA = results.EFA
names, cluster_loadings = zip(*HCA.get_cluster_loading(EFA, rotate=rotate).items())
cluster_DVs = HCA.get_cluster_DVs(inp='EFA%s_%s' % (EFA.get_c(), rotate))
cluster_loadings = list(zip([cluster_DVs[n] for n in names], cluster_loadings))
max_loading = max([max(abs(i[1])) for i in cluster_loadings])
# plot
colors = sns.hls_palette(len(cluster_loadings))
ncols = min(5, len(cluster_loadings))
nrows = ceil(len(cluster_loadings)/ncols)
f, axes = plt.subplots(nrows, ncols,
figsize=(ncols*10,nrows*(8+nrows)),
subplot_kw={'projection': 'polar'})
axes = f.get_axes()
for i, (measures, loading) in enumerate(cluster_loadings):
plot_loadings(axes[i], loading, kind='line', offset=.5,
plot_kws={'alpha': .8, 'c': colors[i]})
axes[i].set_title('Cluster %s' % i, y=1.14, fontsize=25)
# set tick labels
xtick_locs = np.arange(0.0, 2*np.pi, 2*np.pi/len(loading))
axes[i].set_xticks(xtick_locs)
axes[i].set_xticks(xtick_locs+np.pi/len(loading), minor=True)
if i%(ncols*2)==0 or i%(ncols*2)==(ncols-1):
axes[i].set_xticklabels(loading.index, y=.08, minor=True)
# set ylim
axes[i].set_ylim(top=max_loading)
for j in range(i+1, len(axes)):
axes[j].set_visible(False)
plt.subplots_adjust(hspace=.5, wspace=.5)
filename = 'polar_factors_EFA%s_%s.%s' % (c, rotate, ext)
if plot_dir is not None:
save_figure(f, path.join(plot_dir, filename),
{'bbox_inches': 'tight'})
plt.close()
def visualize_importance(importance,
ax,
xticklabels=True,
yticklabels=True,
pad=0,
ymax=None,
legend=True):
"""Plot task loadings on one axis"""
importance_vars = importance[0]
importance_vals = [abs(i) + pad for i in importance[1].T]
plot_loadings(ax,
importance_vals,
kind='line',
offset=.5,
plot_kws={
'alpha': 1,
'linewidth': 3
})
# set up x ticks
xtick_locs = np.arange(0.0, 2 * np.pi, 2 * np.pi / len(importance_vars))
ax.set_xticks(xtick_locs)
ax.set_xticks(xtick_locs + np.pi / len(importance_vars), minor=True)
if xticklabels:
if type(importance_vars[0]) == str:
ax.set_xticklabels(importance_vars, y=.08, minor=True)
else:
ax.set_xticklabels(
['Fac %s' % str(i + 1) for i in importance_vars],
y=.08,
minor=True)
# set up yticks
if ymax:
ax.set_ylim(top=ymax)
ytick_locs = ax.yaxis.get_ticklocs()
new_yticks = np.linspace(0, ytick_locs[-1], 7)
ax.set_yticks(new_yticks)
if yticklabels:
labels = np.round(new_yticks, 2)
replace_dict = {i: '' for i in labels[::2]}
labels = [replace_dict.get(i, i) for i in labels]
ax.set_yticklabels(labels)
if legend:
ax.legend(loc='upper center', bbox_to_anchor=(.5, -.15))
def visualize_importance(importance,
ax,
xticklabels=True,
yticklabels=True,
axes_linewidth=None,
label_size=10,
label_scale=0,
title=None,
ymax=None,
color='k'):
importance_vars = importance[0]
importance_vars = [shortened_factors.get(v, v) for v in importance_vars]
if importance[1] is not None:
importance_vals = [abs(i) for i in importance[1]]
plot_loadings(ax,
importance_vals,
kind='line',
offset=.5,
colors=[color],
plot_kws={
'alpha': 1,
'linewidth': label_size / 4
})
else:
ax.set_yticks([])
ax.grid(linewidth=label_size / 8)
if axes_linewidth:
plt.setp(list(ax.spines.values()), linewidth=axes_linewidth)
# set up x ticks
xtick_locs = np.arange(0.0, 2 * np.pi, 2 * np.pi / len(importance_vars))
ax.set_xticks(xtick_locs)
ax.set_xticks(xtick_locs + np.pi / len(importance_vars), minor=True)
if xticklabels:
if type(importance_vars[0]) != str:
importance_vars = [str(i + 1) for i in importance_vars]
scale = 1 + label_scale
size = ax.get_position().expanded(scale, scale)
ax2 = ax.get_figure().add_axes(size, zorder=2)
for i, var in enumerate(importance_vars):
arc_start = (i + .15) * 2 * np.pi / len(importance_vars)
arc_end = (i + .85) * 2 * np.pi / len(importance_vars)
curve = [
.85 * np.cos(np.linspace(arc_start, arc_end, 100)),
.85 * np.sin(np.linspace(arc_start, arc_end, 100))
]
plt.plot(*curve, alpha=0)
curvetext = CurvedText(
x=curve[0][::-1],
y=curve[1][::-1],
text=var, #'this this is a very, very long text',
va='bottom',
axes=ax2,
fontsize=label_size)
ax2.set_xlim([-1, 1])
ax2.set_ylim([-1, 1])
ax2.axis('off')
if title:
ax.set_title(title, fontsize=label_size * 1.5, y=1.1)
# set up yticks
if len(importance[1]) != 0:
ax.set_ylim(bottom=0)
if ymax is None:
ymax = np.max(importance_vals) * 1.1
ax.set_ylim(top=ymax)
new_yticks = np.linspace(0, ymax, 7)
ax.set_yticks(new_yticks)
if yticklabels:
labels = np.round(new_yticks, 2)
replace_dict = {i: '' for i in labels[::2]}
labels = [replace_dict.get(i, i) for i in labels]
ax.set_yticklabels(labels)