44

I would like to compare a set of distributions of scores (score), grouped by some categories (centrality) and colored by some other (model). I've tried the following with seaborn:

plt.figure(figsize=(14,6))
seaborn.boxplot(x="centrality", y="score", hue="model", data=data, palette=seaborn.color_palette("husl", len(models) +1))
seaborn.despine(offset=10, trim=True)
plt.savefig("/home/i11/staudt/Eval/properties-replication-test.pdf", bbox_inches="tight")

There are some problems I have with this plot:

  • There is a large amount of outliers and I don't like how they are drawn here. Can I remove them? Can I change the appearance to show less clutter? Can I color them at least so that their color matches the box color?
  • The model value original is special because all other distributions should be compared to the distribution of original. This should be visually reflected in the plot. Can I make original the first box of every group? Can I offset or mark it differently somehow? Would it be possible to draw a horizontal line through the median of each original distribution and through the group of boxes?
  • some of the values of score are very small, how to do proper scaling of the y-axis to show them?

enter image description here

EDIT:

Here is an example with a log-scaled y-axis - also not yet ideal. Why do the some boxes seem cut off at the low end?

enter image description here

clstaudt
  • 21,436
  • 45
  • 156
  • 239
  • 1
    The blue outliers are a matplotlib bug, and they can be avoided by updating your seaborn. But I certainly wouldn't just remove the outliers! – mwaskom Feb 01 '16 at 15:52
  • 2
    Have you tried [log-scaling](http://stanford.edu/~mwaskom/software/seaborn/examples/horizontal_boxplot.html) the y variable? – mwaskom Feb 01 '16 at 15:53
  • @mwaskom Yes, but that "compresses" the bars of larger values very much at the top of the plot. Is there a compromise, i.e. log-scaling for just a part of the axis? – clstaudt Feb 02 '16 at 10:15
  • And sometimes the groups are hard to distinguish, e.g. if the boxes do not touch. Is it possible to draw vertical lines separating the groups? – clstaudt Feb 02 '16 at 14:30

2 Answers2

82

Outlier display

You should be able to pass any arguments to seaborn.boxplot that you can pass to plt.boxplot (see documentation), so you could adjust the display of the outliers by setting flierprops. Here are some examples of what you can do with your outliers.

If you don't want to display them, you could do

seaborn.boxplot(x="centrality", y="score", hue="model", data=data,
                showfliers=False)

or you could make them light gray like so:

flierprops = dict(markerfacecolor='0.75', markersize=5,
              linestyle='none')
seaborn.boxplot(x="centrality", y="score", hue="model", data=data,
                flierprops=flierprops)

Order of groups

You can set the order of the groups manually with hue_order, e.g.

seaborn.boxplot(x="centrality", y="score", hue="model", data=data,
                hue_order=["original", "Havel..","etc"])

Scaling of y-axis

You could just get the minimum and maximum values of all y-values and set y_lim accordingly? Something like this:

y_values = data["scores"].values
seaborn.boxplot(x="centrality", y="score", hue="model", data=data,
                y_lim=(np.min(y_values),np.max(y_values)))

EDIT: This last point doesn't really make sense since the automatic y_lim range will already include all the values, but I'm leaving it just as an example of how to adjust these settings. As mentioned in the comments, log-scaling probably makes more sense.

Janosh
  • 3,392
  • 2
  • 27
  • 35
Lisa
  • 3,365
  • 3
  • 19
  • 30
2

It has been a while since this answer has activity, but I'll answer OP's question regarding the weird looking lower-bounds for any people that need help in the future.

Once you set your y-axis to logarithmic scale, it becomes impossible to represent y=0, since log(0) tends to -inf.

Therefore, when the values regarding the lower part of your boxplot are either zero or very close to it the box has that look of seeming to be 'cut in half'.

Needless to say that it's also impossible to represent negative y values in a logarithmic scale.

Dharman
  • 30,962
  • 25
  • 85
  • 135
André Cavalheiro
  • 21
  • 1