
上QQ阅读APP看书,第一时间看更新
Graphing Anscombe's quartet
Anscombe's quartet is a classic example that illustrates why visualizing data is important. The quartet consists of four datasets with similar statistical properties. Each dataset has a series of x values and dependent y values. We will tabulate these metrics in an IPython notebook. However, if you plot the datasets, they look surprisingly different compared to each other.
How to do it...
For this recipe, you need to perform the following steps:
- Start with the following imports:
import pandas as pd import seaborn as sns import matplotlib.pyplot as plt import matplotlib as mpl from dautil import report from dautil import plotting import numpy as np from tabulate import tabulate
- Define the following function to compute the mean, variance, and correlation of
x
andy
within a dataset, the slope, and the intercept of a linear fit for each of the datasets:df = sns.load_dataset("anscombe") agg = df.groupby('dataset')\ .agg([np.mean, np.var])\ .transpose() groups = df.groupby('dataset') corr = [g.corr()['x'][1] for _, g in groups] builder = report.DFBuilder(agg.columns) builder.row(corr) fits = [np.polyfit(g['x'], g['y'], 1) for _, g in groups] builder.row([f[0] for f in fits]) builder.row([f[1] for f in fits]) bottom = builder.build(['corr', 'slope', 'intercept']) return df, pd.concat((agg, bottom))
- The following function returns a string, which is partly Markdown, partly restructured text, and partly HTML, because core Markdown does not officially support tables:
def generate(table): writer = report.RSTWriter() writer.h1('Anscombe Statistics') writer.add(tabulate(table, tablefmt='html', floatfmt='.3f')) return writer.rst
- Plot the data and corresponding linear fits with the Seaborn
lmplot()
function:def plot(df): sns.set(style="ticks") g = sns.lmplot(x="x", y="y", col="dataset", hue="dataset", data=df, col_wrap=2, ci=None, palette="muted", size=4, scatter_kws={"s": 50, "alpha": 1}) plotting.embellish(g.fig.axes)
- Display a table with statistics, as follows:
df, table = aggregate() from IPython.display import display_markdown display_markdown(generate(table), raw=True)
The following table shows practically identical statistics for each dataset (I modified the
custom.css
file in my IPython profile to get the colors): - The following lines plot the datasets:
%matplotlib inline plot(df)
Refer to the following plot for the end result:

A picture says more than a thousand words. The source code is in the anscombe.ipynb
file in this book's code bundle.
See also
- The Anscombe's quartet Wikipedia page at https://en.wikipedia.org/wiki/Anscombe%27s_quartet (retrieved July 2015)
- The seaborn documentation for the
lmplot()
function at https://web.stanford.edu/~mwaskom/software/seaborn/generated/seaborn.lmplot.html (retrieved July 2015)