A random variable is a function, not a wild number
Despite the name, a random variable is a measurable map from outcomes to numbers. Heads might become one and tails zero. The distribution that map pushes forward forgets the underlying space—and two variables can share a law yet remain independent or entangled.
Mathematically, a random variable formalizes a quantity that depends on chance events. The phrase does not literally mean randomness plus variability; it names a function whose domain is a sample space of outcomes and whose range is a measurable space, often a subset of the reals. Chance may mean a die roll or measurement error; philosophy of probability can be messy, but the analysis sits on axiomatic measure theory.
Formally it is a measurable function from a probability space to a measurable space. The pushforward measure is the variable's distribution—a probability measure on possible values. Two variables may share identical distributions yet differ, for example by being independent. Discrete variables take countable values and admit mass functions; absolutely continuous ones on intervals admit densities, with single points carrying probability zero. Not every continuous variable is absolutely continuous.
Every random variable has a cumulative distribution function giving the chance of being at most a threshold. Statistics traditionally stresses real-valued cases so expectation, variance, and moments make sense. Broader random elements cover Booleans, categories, vectors, matrices, graphs, shapes, and functions—useful in machine learning and discrete mathematics. A random word may be an index into a vocabulary or a one-hot indicator vector.
Call a time-indexed family of random values a stochastic process; spread the same idea over space and you have a random field. A random variate is one realized outcome. George Mackey credited Pafnuty Chebyshev as first to think systematically in these terms. Asking how likely X equals two means measuring the set of outcomes mapped to two—then collecting those chances builds the distribution that forgets the original space's details.
Source: Random variable
