probability-calculator 0.3.0

Calculate with and analyze probability densities.

probability_calculator

Calculate with and analyze random variables / probability densities.

Usage

Install the package, e.g.: pip install probability_calculator.

Initialization and plotting

The package provides simple ways to define random variables with discrete outcomes, for instance a die:

from probability_calculator import FairDie

density = FairDie(6) # initialize a fair die with 6 sides
fig, ax = density.plot_outcomes() # plot the outcomes using matplotlib

For the general case, the class RandomVariables can be used:

from probability_calculator import RandomVariable
from fractions import Fraction

density = RandomVariable(outcomes=[
    { "value": 0, "p": Fraction(2, 10) },
    { "value": 1, "p": Fraction(3, 10) },
    { "value": Fraction(5, 2), "p": Fraction(1, 10) },
    { "value": 3, "p": Fraction(4, 10) },
]) # initialize a discrete density with 4 different outcomes
fig, ax = density.plot_outcomes() # plot the density using matplotlib

Note that only integers and fractions can be used to assure exact calculations.

Combine random variables

The discrete density of throwing a die two times can be modelled by adding the random variable with itself:

density_for_one_throw = FairDie(6)
density_sum_of_two_throws = density_for_one_throw + density_for_one_throw # same as density_for_one_throw*2
fig, ax = density_sum_of_two_throws.plot_outcomes()

Note that the operations (+ and *) on random variables are always assuming independed random variables, even when using the same variable multiple times.

Limitations

Continuous densities / random variables are not supported at the moment.

Adding many random variables together might get stuck due to a lot of possible outcomes. In general, adding 10 densities with 10 outcomes each lead to $10^{10}$ outcomes. However, simple cases like the die work, so even Die(10)*100 is no problem.

Contributing

We greatly appreciate fixes and new features for probability_calculator. All contributions to this project should be sent as pull requests on github.

License

Apache License 2.0