问题
Can anyone give me a way to do a qq plot in Seaborn as a test for normality of data? Or failing that, at least in matplotlib.
Thanks in advance
回答1:
After reading the wikipedia article, I understand that the Q-Q plot is a plot of the quantiles of two distributions against each other.
numpy.percentile allows to obtain the percentile of a distribution. Hence you can call numpy.percentile on each of the distributions and plot the results against each other.
import numpy as np
import matplotlib.pyplot as plt
a = np.random.normal(5,5,250)
b = np.random.rayleigh(5,250)
percs = np.linspace(0,100,21)
qn_a = np.percentile(a, percs)
qn_b = np.percentile(b, percs)
plt.plot(qn_a,qn_b, ls="", marker="o")
x = np.linspace(np.min((qn_a.min(),qn_b.min())), np.max((qn_a.max(),qn_b.max())))
plt.plot(x,x, color="k", ls="--")
plt.show()
回答2:
I'm not sure if this still recent, but I notice that neither of the answers really addresses the question, which asks how to do qq-plots with scipy and seaborn, but doesn't mention statsmodels. In fact, qq-plots are available in scipy under the name probplot:
from scipy import stats
import seaborn as sns
stats.probplot(x, plot=sns.mpl.pyplot)
The plot argument to probplot can be anything that has a plot method and a text method. Probplot is also quite flexible about the kinds of theoretical distributions it supports.
回答3:
Statsmodels.api has a plotter for this job, check out statsmodels.api.qqplot().
Here's a couple of examples using same data as above.
import numpy as np
import matplotlib.pyplot as plt
import statsmodels.api as sm
a = np.random.normal(5,5,250)
sm.qqplot(a)
plt.show()
SO won't let me post images yet. So try it out. This one shows a normal distribution plotted against a normal distribution resulting in (pretty darn) straight line.
b = np.random.rayleigh(5,250)
sm.qqplot(b)
This one shows raleigh dist plotted against normal dist. Result is slightly concave curve.
来源：https://stackoverflow.com/questions/46935289/quantile-quantile-plot-using-seaborn-and-scipy