ML Wiki
Machine Learning Wiki - A collection of ML concepts, algorithms, and resources.

Chain and Sum Rules in Probability

Addition Theorem of Probabilities

The sum $A + B$ of two events $A$ and $B$ is the event consisting of the occurrence of event $A$ or event $B$.

Theorem. The probability of occurrence of one of two mutually exclusive events equals the sum of their probabilities:

$P(A + B) = P(A) + P(B)$

Proof: $n$ - total number of outcomes, $m_a$ - outcomes favorable to $A$, $m_b$ - outcomes favorable to $B$

$P(A + B) = \frac{m_a + m_b}{n} = \frac{m_a}{n} + \frac{m_b}{n} = P(A) + P(B)$

Corollaries

  • The sum of probabilities of all events $A_i \in \Omega$ forming a complete group of events equals one. $P(A_1) + … + P(A_n) = 1$
  • The sum of the probability of event $A$ and its complementary event $\bar{A}$ equals one, since $A$ and $\bar{A}$ form a complete group of events. $P(A) + P(\bar{A}) = 1$

Addition Theorem for Compatible Events

Two events are called compatible if the occurrence of one does not exclude the occurrence of the other in the same trial.

Theorem. The probability of occurrence of at least one of two compatible events equals the sum of their probabilities minus the probability of their joint occurrence:

$P(A + B) = P(A) + P(B) - P(AB)$

Proof:

  • $A + B$ occurs if $A\bar{B}$, $\bar{A}B$, or $AB$ occurs. Since these events are mutually exclusive, by the addition theorem we have $P(A + B) = P(A\bar{B}) + P(\bar{A}B) + P(AB)$ (*)
  • $A$ occurs if either $AB$ or $A\bar{B}$ occurs. By the addition theorem, $P(A) = P(A\bar{B}) + P(AB)$ or $P(A\bar{B}) = P(A) - P(AB)$ (**)
  • Similarly, $B$ occurs if either $AB$ or $\bar{A}B$ occurs. That is, $P(B) = P(\bar{A}B) + P(AB)$ or $P(\bar{A}B) = P(B) - P(AB)$ (***)
  • Substituting (**) and (***) into (*), we get $P(A + B) = P(A) + P(B) - P(AB)$

Multiplication Theorem of Probabilities

The product of events $A$ and $B$ is the event $A \cdot B$ consisting of the joint occurrence of these events.

Example:

  • $A$ - the part is functional
  • $B$ - the part is painted
  • $A \cdot B$ - the part is functional and painted

Theorem. Consider two events $A$ and $B$. We know $P(A)$ and $P(B\mid A)$. How do we find the probability that both $A$ and $B$ occur? $P(A \cdot B) = P(A) \cdot P(B \mid A)$ - by the definition of conditional probability. For independent events the multiplication theorem becomes

$P(A \cdot B) = P(A) \cdot P(B)$

See also

Sources

  • Gmurman V.E., Probability Theory and Mathematical Statistics – 9th edition. Moscow: Vysshaya Shkola, 2003.