The negative binomial distribution is used for independent Bernoulli trials and measures the number of tries (X=k) that are needed before a specified number of successes (r) occur. An example would be the number of coin tosses it would take to obtain five heads. The PMF is given as follows:
The expectation and variance are given respectively by the following expressions:
We can see that the negative binomial is a generalization of the geometric distribution, with the geometric distribution being a special case of the negative binomial, where r=1.
The code and plot are shown as follows:
In [189]: from scipy.stats import nbinom from matplotlib import colors clrs = matplotlib.rcParams['axes.color_cycle'] x = np.arange(0,11) n_vals = [0.1,1,3,6] p=0.5 for n, clr in zip(n_vals, clrs): rv = nbinom(n,p) plt.plot(x,rv.pmf(x), label="$n$=" + str(n), color=clr) plt.legend() plt.title("Negative Binomial Distribution PMF") plt.ylabel("PMF at $x$") plt.xlabel("$x$")
The following is the output:
Negative binomial distribution