def test_violinplot_beanplot():
# Test violinplot and beanplot with the same dataset.
data = anes96.load_pandas()
party_ID = np.arange(7)
labels = ["Strong Democrat", "Weak Democrat", "Independent-Democrat",
"Independent-Independent", "Independent-Republican",
"Weak Republican", "Strong Republican"]
age = [data.exog['age'][data.endog == id] for id in party_ID]
fig = plt.figure()
ax = fig.add_subplot(111)
violinplot(age, ax=ax, labels=labels,
plot_opts={'cutoff_val':5, 'cutoff_type':'abs',
'label_fontsize':'small',
'label_rotation':30})
plt.close(fig)
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age, ax=ax, labels=labels,
plot_opts={'cutoff_val':5, 'cutoff_type':'abs',
'label_fontsize':'small',
'label_rotation':30})
plt.close(fig)
def beanplot_grid(vals, names, fig=None, title=None):
from matplotlib import pyplot as plt
from mpl_toolkits.axes_grid1 import ImageGrid
from statsmodels.graphics.boxplots import beanplot
if fig is None:
fig = plt.figure()
if title:
fig.suptitle(title)
n = vals.shape[1]
grid = ImageGrid(fig,
111,
nrows_ncols=(n, n),
aspect=False,
share_all=True,
label_mode='L')
for i in range(n * n):
grid[i].set_visible(False)
for j, i in itertools.combinations_with_replacement(range(n), 2):
idx = i * n + j
grid[idx].set_visible(True)
data = vals[:, i, j]
if np.ptp(data) == 0:
grid[idx].hlines(data[0], .85, 1.15, lw=0.5, color='k')
else:
beanplot([data],
labels=[''],
ax=grid[idx],
plot_opts={'cutoff': True})
# TODO: actually show the names...
return grid
def plot_aucs(aucs, ylabel=None, names=None, rotation=90, ha='center'):
import matplotlib.pyplot as plt
if names is None:
names = KEY_NAMES
names, aucs = list(zip(*sorted((names[k], v) for k, v in aucs.items())))
if all(a.size == 1 for a in aucs):
plt.plot(aucs, linestyle='None', marker='o')
indices = np.arange(len(names))
else:
try:
from statsmodels.graphics.boxplots import beanplot
except ImportError:
plt.boxplot(aucs)
else:
# jiggle anything that's all exactly one value, to avoid singularity
aucs = [
grp if len(set(grp)) > 1 else list(grp) + [grp[0] + .01]
for grp in aucs
]
beanplot(aucs, ax=plt.gca(), plot_opts={'cutoff': True})
indices = np.arange(len(names)) + 1
plt.xticks(indices, names, rotation=rotation, ha=ha)
plt.xlim(indices[0] - .5, indices[-1] + .5)
bot, top = plt.ylim()
if bot < 0 < top:
plt.hlines(0, *plt.xlim(), color='k')
if ylabel:
plt.ylabel(ylabel)
def plot_lnprob(tup):
"""
Bean plot of prior vs. posterior log-probabilities
Parameters
----------
tup: tuple of arrays
Tuple (lnprob_prior, lnprob_post) of log-probabilities
Returns
-------
fig: matplotlib figure
The figure with the plot
"""
try:
from statsmodels.graphics.boxplots import beanplot
fig = pl.figure()
ax = fig.gca()
beanplot(tup, ax=ax, jitter=False, labels=("prior","posterior"),
plot_opts={"bean_mean_color":"c",
"bean_median_color":"m",
"violin_fc":"w",
"violin_alpha":1})
ax.set_xlabel("prior vs. posterior")
ax.set_ylabel("log-probability")
return fig
except ImportError:
print "Consider installing StatsModels package for bean plots."
def beanplot_grid(vals, names, fig=None, title=None):
from matplotlib import pyplot as plt
from mpl_toolkits.axes_grid1 import ImageGrid
from statsmodels.graphics.boxplots import beanplot
if fig is None:
fig = plt.figure()
if title:
fig.suptitle(title)
n = vals.shape[1]
grid = ImageGrid(fig, 111, nrows_ncols=(n, n), aspect=False,
share_all=True, label_mode='L')
for i in range(n * n):
grid[i].set_visible(False)
for j, i in itertools.combinations_with_replacement(range(n), 2):
idx = i * n + j
grid[idx].set_visible(True)
data = vals[:, i, j]
if np.ptp(data) == 0:
grid[idx].hlines(data[0], .85, 1.15, lw=0.5, color='k')
else:
beanplot([data], labels=[''], ax=grid[idx],
plot_opts={'cutoff': True})
# TODO: actually show the names...
return grid
def plot_lnprob(tup):
"""
Bean plot of prior vs. posterior log-probabilities
Parameters
----------
tup: tuple of arrays
Tuple (lnprob_prior, lnprob_post) of log-probabilities
Returns
-------
fig: matplotlib figure
The figure with the plot
"""
try:
from statsmodels.graphics.boxplots import beanplot
fig = pl.figure()
ax = fig.gca()
beanplot(tup,
ax=ax,
jitter=False,
labels=("prior", "posterior"),
plot_opts={
"bean_mean_color": "c",
"bean_median_color": "m",
"violin_fc": "w",
"violin_alpha": 1
})
ax.set_xlabel("prior vs. posterior")
ax.set_ylabel("log-probability")
return fig
except ImportError:
print "Consider installing StatsModels package for bean plots."
def test_violinplot_beanplot(close_figures):
# Test violinplot and beanplot with the same dataset.
data = anes96.load_pandas()
party_ID = np.arange(7)
labels = ["Strong Democrat", "Weak Democrat", "Independent-Democrat",
"Independent-Independent", "Independent-Republican",
"Weak Republican", "Strong Republican"]
age = [data.exog['age'][data.endog == id] for id in party_ID]
fig = plt.figure()
ax = fig.add_subplot(111)
violinplot(age, ax=ax, labels=labels,
plot_opts={'cutoff_val':5, 'cutoff_type':'abs',
'label_fontsize':'small',
'label_rotation':30})
fig = plt.figure()
ax = fig.add_subplot(111)
violinplot(age, ax=ax, labels=labels,
plot_opts={'cutoff_val':5, 'cutoff_type':'abs',
'label_fontsize':'small',
'label_rotation':30,
'bw_factor':.2})
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age, ax=ax, labels=labels,
plot_opts={'cutoff_val':5, 'cutoff_type':'abs',
'label_fontsize':'small',
'label_rotation':30})
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age, ax=ax, labels=labels, jitter=True,
plot_opts={'cutoff_val': 5, 'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30})
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age, ax=ax, labels=labels, jitter=True, side='right',
plot_opts={'cutoff_val': 5, 'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30})
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age, ax=ax, labels=labels, jitter=True, side='left',
plot_opts={'cutoff_val': 5, 'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30})
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age, ax=ax, labels=labels,
plot_opts={'bean_legend_text': 'text'})
def test_beanplot(age_and_labels, close_figures):
age, labels = age_and_labels
fig, ax = plt.subplots(1, 1)
beanplot(age,
ax=ax,
labels=labels,
plot_opts={
'cutoff_val': 5,
'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30
})
def make_35_binary(indata, cyto, ax, confounders):
valid_mask = (indata['Age'] >= 50) | (indata['Age'] <= 35)
eqn, tmp_cyto = extract_cols(indata[valid_mask], cyto,
['OlderThan50']+confounders)
y, X = dmatrices(eqn, tmp_cyto, return_type='dataframe')
model = sm.OLS(y, X).fit()
boxes = [tmp_cyto['y'][tmp_cyto['OlderThan50']], tmp_cyto['y'][~tmp_cyto['OlderThan50']]]
if ax is not None:
try:
beanplot(boxes, labels=['>50', '<35'], ax=ax)
except:
ax.boxplot(boxes)
return model, model.f_pvalue, model.pvalues['OlderThan50[T.True]'], model.params['OlderThan50[T.True]']
def test_violinplot_beanplot():
"""Test violinplot and beanplot with the same dataset."""
data = anes96.load_pandas()
party_ID = np.arange(7)
labels = [
"Strong Democrat", "Weak Democrat", "Independent-Democrat",
"Independent-Independent", "Independent-Republican", "Weak Republican",
"Strong Republican"
]
age = [data.exog['age'][data.endog == id] for id in party_ID]
fig = plt.figure()
ax = fig.add_subplot(111)
violinplot(age,
ax=ax,
labels=labels,
plot_opts={
'cutoff_val': 5,
'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30
})
plt.close(fig)
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age,
ax=ax,
labels=labels,
plot_opts={
'cutoff_val': 5,
'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30
})
plt.close(fig)
def make_50_binary(indata, cyto, ax, confounders):
eqn, tmp_cyto = extract_cols(indata, cyto, ['OlderThan50']+confounders)
y, X = dmatrices(eqn, tmp_cyto, return_type='dataframe')
try:
model = sm.OLS(y, X).fit()
except:
print indata
print confounders
print eqn
print tmp_cyto
print X
print y
raise TypeError
if ax is not None:
boxes = [tmp_cyto['y'][tmp_cyto['OlderThan50']], tmp_cyto['y'][~tmp_cyto['OlderThan50']]]
try:
beanplot(boxes, labels=['>50', '<50'], ax=ax)
except:
ax.boxplot(boxes)
return model, model.f_pvalue, model.pvalues['OlderThan50[T.True]'], model.params['OlderThan50[T.True]']
def test_beanplot_legend_text(age_and_labels, close_figures):
age, labels = age_and_labels
fig, ax = plt.subplots(1, 1)
beanplot(age, ax=ax, labels=labels, plot_opts={'bean_legend_text': 'text'})
def test_violinplot_beanplot(close_figures):
# Test violinplot and beanplot with the same dataset.
data = anes96.load_pandas()
party_ID = np.arange(7)
labels = [
"Strong Democrat", "Weak Democrat", "Independent-Democrat",
"Independent-Independent", "Independent-Republican", "Weak Republican",
"Strong Republican"
]
age = [data.exog['age'][data.endog == id] for id in party_ID]
fig = plt.figure()
ax = fig.add_subplot(111)
violinplot(age,
ax=ax,
labels=labels,
plot_opts={
'cutoff_val': 5,
'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30
})
fig = plt.figure()
ax = fig.add_subplot(111)
violinplot(age,
ax=ax,
labels=labels,
plot_opts={
'cutoff_val': 5,
'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30,
'bw_factor': .2
})
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age,
ax=ax,
labels=labels,
plot_opts={
'cutoff_val': 5,
'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30
})
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age,
ax=ax,
labels=labels,
jitter=True,
plot_opts={
'cutoff_val': 5,
'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30
})
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age,
ax=ax,
labels=labels,
jitter=True,
side='right',
plot_opts={
'cutoff_val': 5,
'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30
})
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age,
ax=ax,
labels=labels,
jitter=True,
side='left',
plot_opts={
'cutoff_val': 5,
'cutoff_type': 'abs',
'label_fontsize': 'small',
'label_rotation': 30
})
fig = plt.figure()
ax = fig.add_subplot(111)
beanplot(age, ax=ax, labels=labels, plot_opts={'bean_legend_text': 'text'})
from statsmodels.graphics.boxplots import beanplot
labels = ['Old', 'Young', 'Pooled']
for cyto in cytos:
bins = []
labs = []
for gname, gmask in comp:
td = sf_data[cyto][gmask].dropna()
if len(td) > 2:
bins.append(td.copy())
labs.append(gname)
if len(bins) > 1:
fig = plt.figure()
ax = plt.subplot(111)
beanplot(bins, labels=labs, ax=ax)
ax.set_ylim([0, ax.get_ylim()[1]])
ax.set_ylabel(cyto)
ax.set_title(cyto)
plt.savefig('/home/will/KrebbsData/draft_figures/violin_plot_%s.png' % cyto)
plt.close()
# <codecell>
print res[(res['Pval'] < 0.1)][['Cyto','Group1', 'G1mean','Group2','G2mean','Pval']]
# <codecell>
wanted_cl = set(['Ect1', 'End1', 'VK2'])
cl_data = data[data['CellLine'].map(lambda x: x in wanted_cl)]
