Boxplot

Shows the use of the functional Boxplot applied to the Canadian Weather dataset.

# Author: Amanda Hernando Bernabé
# License: MIT

# sphinx_gallery_thumbnail_number = 2

from skfda import datasets
from skfda.exploratory.depth import ModifiedBandDepth, IntegratedDepth
from skfda.exploratory.visualization import Boxplot

import matplotlib.pyplot as plt
import numpy as np

First, the Canadian Weather dataset is downloaded from the package ‘fda’ in CRAN. It contains a FDataGrid with daily temperatures and precipitations, that is, it has a 2-dimensional image. We are interested only in the daily average temperatures, so we will use the first coordinate.

X, y = datasets.fetch_weather(return_X_y=True, as_frame=True)
fd = X.iloc[:, 0].values
fd_temperatures = fd.coordinates[0]

The data is plotted to show the curves we are working with. They are divided according to the target. In this case, it includes the different climates to which the weather stations belong to.

# Each climate is assigned a color. Defaults to grey.
colormap = plt.cm.get_cmap('seismic')
label_names = y.values.categories
nlabels = len(label_names)
label_colors = colormap(np.arange(nlabels) / (nlabels - 1))

fd_temperatures.plot(group=y.values.codes,
                     group_colors=label_colors,
                     group_names=label_names)

/home/docs/checkouts/readthedocs.org/user_builds/fda/checkouts/latest/examples/plot_boxplot.py:37: MatplotlibDeprecationWarning: The get_cmap function was deprecated in Matplotlib 3.7 and will be removed two minor releases later. Use ``matplotlib.colormaps[name]`` or ``matplotlib.colormaps.get_cmap(obj)`` instead.
  colormap = plt.cm.get_cmap('seismic')

<Figure size 640x480 with 1 Axes>

We instantiate a Boxplot object with the data, and we call its plot() function to show the graph.

By default, only the part of the outlier curves which falls out of the central regions is plotted. We want the entire curve to be shown, that is why the show_full_outliers parameter is set to True.

fdBoxplot = Boxplot(fd_temperatures)
fdBoxplot.show_full_outliers = True

fdBoxplot.plot()

<Figure size 640x480 with 1 Axes>

We can observe in the boxplot the median in black, the central region (where the 50% of the most centered samples reside) in pink and the envelopes and vertical lines in blue. The outliers detected, those samples with at least a point outside the outlying envelope, are represented with a red dashed line. The colors can be customized.

The outliers are shown below with respect to the other samples.

color = 0.3
outliercol = 0.7

fd_temperatures.plot(group=fdBoxplot.outliers.astype(int),
                     group_colors=colormap([color, outliercol]),
                     group_names=["nonoutliers", "outliers"])

<Figure size 640x480 with 1 Axes>

The curves pointed as outliers are are those curves with significantly lower values than the rest. This is the expected result due to the depth measure used, IntegratedDepth(), which ranks the samples according to their magnitude.

The Boxplot object admits any depth measure defined or customized by the user. Now the call is done with the ModifiedBandDepth and the factor is reduced in order to designate some samples as outliers (otherwise, with this measure and the default factor, none of the curves are pointed out as outliers). We can see that the outliers detected belong to the Pacific and Arctic climates which are less common to find in Canada. As a consequence, this measure detects better shape outliers compared to the previous one.

fdBoxplot = Boxplot(
    fd_temperatures, depth_method=ModifiedBandDepth(), factor=0.4)
fdBoxplot.show_full_outliers = True

fdBoxplot.plot()

<Figure size 640x480 with 1 Axes>

Another functionality implemented in this object is the enhanced functional boxplot, which can include other central regions, apart from the central or 50% one.

In the following instantiation, the IntegratedDepth() is used and the 25% and 75% central regions are specified.

fdBoxplot = Boxplot(fd_temperatures, depth_method=IntegratedDepth(),
                    prob=[0.75, 0.5, 0.25])
fdBoxplot.plot()

<Figure size 640x480 with 1 Axes>

The above two lines could be replaced just by fdBoxplot inside a notebook since the default representation of the Boxplot is the image of the plot.