rv_discrete#

Abstraction: SciPy base class for subclassing custom discrete probability distributions

Key points:

• scipy.stats.rv_discrete is a generic base class; subclass it and override _pmf (or _cdf) to define a custom discrete distribution
• Can also instantiate directly with values=(xk, pk) to create a finite discrete distribution from explicit probability masses summing to 1
• Provides full statistical method suite: rvs, pmf, cdf, sf, ppf, isf, stats, entropy, mean, var, std, median, interval
• Key differences from rv_continuous: support is integers; uses probability mass function (pmf) not pdf; no scale parameter; default _cdf assumes bounded-below support
• Deepcopy/pickling isolates the RNG state, so np.random.seed no longer controls copied distribution instances

Connections: Scipy · Numpy · Probability Distributions · Scientific Computing · Python Data Science

Source: http://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.rv_discrete.html