Skip to content

Updated violin plot example as per suggestions in issue #7251 #7360

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 9 commits into from
Oct 31, 2016
Merged

Updated violin plot example as per suggestions in issue #7251 #7360

Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
08db92d
Updated example as per suggestions in issue #7251
DrNightmare Oct 29, 2016
becbfa4
Fixed ordering of parameters to np.clip()
DrNightmare Oct 29, 2016
14cd6cf
Made variable names more meaningful
DrNightmare Oct 30, 2016
03946d4
Replaced use of zip with call to transpose
DrNightmare Oct 30, 2016
8044e9d
Changed variable 'tmp' to 'quartiles' and slightly altered transpose …
DrNightmare Oct 30, 2016
88a8c2e
Fixed recalculation of interquartile ranges
DrNightmare Oct 30, 2016
62c1716
Replaced plot() calls with scatter() and vlines() calls
DrNightmare Oct 30, 2016
9f4bbdd
Removed unused variable and comments
DrNightmare Oct 31, 2016
b3ce73b
Removed unnecessary comments
DrNightmare Oct 31, 2016
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Next Next commit
Updated example as per suggestions in issue #7251
  • Loading branch information
@DrNightmare
DrNightmare committed Oct 29, 2016
commit 08db92d1ed16f61148323e2656168611880f0de9
107 changes: 47 additions & 60 deletions examples/statistics/customized_violin_demo.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,87 +18,74 @@
import numpy as np


# functions to calculate percentiles and adjacent values
def percentile(vals, p):
N = len(vals)
n = p*(N+1)
k = int(n)
d = n-k
if k <= 0:
return vals[0]
if k >= N:
return vals[N-1]
return vals[k-1] + d*(vals[k] - vals[k-1])


def adjacent_values(vals):
q1 = percentile(vals, 0.25)
q3 = percentile(vals, 0.75)
iqr = q3 - q1 # inter-quartile range

q1, q3 = np.percentile(vals, [25, 75])
# inter-quartile range iqr
iqr = q3 - q1
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What about removing this comment and instead changing to clearer variable names:
interquartile_range = q3 - q1

# upper adjacent values
uav = q3 + iqr * 1.5
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

same, change to
upper_adajcent_values = q3 - interquartile_range*1.5

if uav > vals[-1]:
uav = vals[-1]
if uav < q3:
uav = q3

uav = np.clip(uav, q3, vals[-1])
# lower adjacent values
lav = q1 - iqr * 1.5
if lav < vals[0]:
lav = vals[0]
if lav > q1:
lav = q1
lav = np.clip(lav, q1, vals[0])
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think you might have these two backwards.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could you please comment a bit more on what you mean by backwards?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

q1 is the maximum and vals[0] is the minimum.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am sorry, I do have them backwards. Changing that now

return [lav, uav]


def set_axis_style(ax, labels):
ax.get_xaxis().set_tick_params(direction='out')
ax.xaxis.set_ticks_position('bottom')
ax.set_xticks(np.arange(1, len(labels) + 1))
ax.set_xticklabels(labels)
ax.set_xlim(0.25, len(labels) + 0.75)
ax.set_xlabel('Sample name')


# create test data
np.random.seed(123)
dat = [np.random.normal(0, std, 100) for std in range(1, 5)]
lab = ['A', 'B', 'C', 'D'] # labels
med = [] # medians
iqr = [] # inter-quantile ranges
avs = [] # upper and lower adjacent values
for arr in dat:
sarr = sorted(arr)
med.append(percentile(sarr, 0.5))
iqr.append([percentile(sarr, 0.25), percentile(sarr, 0.75)])
avs.append(adjacent_values(sarr))

# plot the violins
fig, (ax1, ax2) = plt.subplots(nrows=1, ncols=2, figsize=(9, 4),
sharey=True)
_ = ax1.violinplot(dat)
parts = ax2.violinplot(dat, showmeans=False, showmedians=False,
showextrema=False)
dat = [sorted(np.random.normal(0, std, 100)) for std in range(1, 5)]

fig, (ax1, ax2) = plt.subplots(nrows=1, ncols=2, figsize=(9, 4), sharey=True)

# plot the default violin
ax1.set_title('Default violin plot')
ax2.set_title('Customized violin plot')
ax1.set_ylabel('Observed values')
ax1.violinplot(dat)

# plot whiskers as thin lines, quartiles as fat lines,
# and medians as points
for i in range(len(med)):
# whiskers
ax2.plot([i + 1, i + 1], avs[i], '-', color='black', linewidth=1)
ax2.plot([i + 1, i + 1], iqr[i], '-', color='black', linewidth=5)
ax2.plot(i + 1, med[i], 'o', color='white',
markersize=6, markeredgecolor='none')
# customized violin
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

stray redundant comment

Copy link
Contributor Author

@DrNightmare DrNightmare Oct 31, 2016 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hmm, right. Guess the title we set for the axes are self explanatory.
Updated with comments removed @story645

ax2.set_title('Customized violin plot')
parts = ax2.violinplot(
dat, showmeans=False, showmedians=False,
showextrema=False)

# customize colors
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

probably also unnecessary comment

for pc in parts['bodies']:
pc.set_facecolor('#D43F3A')
pc.set_edgecolor('black')
pc.set_alpha(1)

ax1.set_ylabel('Observed values')
# medians
med = [np.percentile(sarr, 50) for sarr in dat]
# inter-quartile ranges
iqr = [[np.percentile(sarr, 25), np.percentile(sarr, 75)] for sarr in dat]
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think you can use the axis parameter and multiple q to compute all three of these at the same time without the list comprehension.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

To compute all of those in one call (without list comprehensions), I could do something like:

for sarr in dat:
    median, q1, q3 = np.percentile(sarr, [50, 25, 75])
    med.append(median)
    iqr.append([q1, q3])
    avs.append(adjacent_values(sarr))

Is that fine?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think you can pass the entire array and use the axis parameter to get it to work on the right dimension. avs would probably need to stay as the list comprehension.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Got it.
I could replace
med = [np.percentile(sarr, 50) for sarr in dat]
with
med = np.percentile(dat, 50, axis=1)

Since iqr is a list of tuples (or rather, list of lists of size 2), I am unable to see how we could do all 3 percentiles with a single call. I could still replace
iqr = [[np.percentile(sarr, 25), np.percentile(sarr, 75)] for sarr in dat]
with
iqr = zip(*(np.percentile(dat, [25, 75], axis=1)))
if that is fine.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No need for the zip; just use .T to swap the axes. You could do it in one call with a temporary variable, but I'm not sure it's any better looking:

tmp = np.percentile(dat, [25, 50, 75], axis=1))).T
med = tmp[:, 1]
iqr = tmp[:, [0, 2]]

# upper and lower adjacent values
avs = [adjacent_values(sarr) for sarr in dat]
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

since this is the whiskers, why not change the variable name to whiskers? and again, then removing the comment


# plot whiskers as thin lines, quartiles as fat lines,
# and medians as points
for i, median in enumerate(med):
# whiskers
ax2.plot([i + 1, i + 1], avs[i], '-', color='black', linewidth=1)
# quartiles
ax2.plot([i + 1, i + 1], iqr[i], '-', color='black', linewidth=5)
# medians
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

doesn't need this comment 'cause clear variable name. Also, still in favor of vlines for this (like in #6814) because I think that's a clearer indication of what's being plotted.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@story645 👍 for better variable names, I have made changes and pushed a commit. I also intend to include the change to vlines in a separate commit, if we go ahead with that.

ax2.plot(
i + 1, median, 'o', color='white',
markersize=6, markeredgecolor='none')

# set style for the axes
labels = ['A', 'B', 'C', 'D'] # labels
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

variable name = comment

for ax in [ax1, ax2]:
ax.get_xaxis().set_tick_params(direction='out')
ax.xaxis.set_ticks_position('bottom')
ax.set_xticks(np.arange(1, len(lab) + 1))
ax.set_xticklabels(lab)
ax.set_xlim(0.25, len(lab) + 0.75)
ax.set_xlabel('Sample name')
set_axis_style(ax, labels)

plt.subplots_adjust(bottom=0.15, wspace=0.05)

plt.show()