Mathematics
The Law of Total Probability and Partitioning Events
Quick fact
The law of total probability lets you find the probability of an event by averaging its conditional probabilities over a partition of the sample space, weighted by the probabilities of the partition parts. For example, the overall probability of a positive test result is the sum of true positive and false positive rates, weighted by how common the condition is.
Why this is interesting
You know the probability it rains on a random day, but have you ever wondered how to compute it when you only know the chances of rain given the season? The law of total probability is the key.
Read the full explanation
Understanding The Law of Total Probability and Partitioning Events
Imagine you want to know the probability that a random person is left-handed, but you only have data for different age groups. You can break the population into age groups (children, adults, seniors) that are mutually exclusive and cover everyone. For each group, you know the probability of being left-handed given that group, and you know the probability a person is in that group. To get the overall probability, you multiply each conditional probability by the group's size and sum them up. This is exactly the law of total probability: it turns a hard calculation into several easier ones.
A deeper explanation
The law of total probability is a consequence of the axioms of probability and the additivity of mutually exclusive events. Formally, if events B₁, B₂, ..., Bₙ form a partition of the sample space (they are pairwise disjoint and their union is the whole space), then for any event A: P(A) = Σᵢ P(A|Bᵢ)P(Bᵢ). This works because A can be written as the disjoint union of A∩Bᵢ, and the probability of a union of disjoint events is the sum of their probabilities. The law is a cornerstone because it lets you compute marginal probabilities from conditional ones, and it is the foundation for Bayes' theorem, which reverses the conditioning. Its power is evident in fields like medicine (calculating overall disease prevalence from risk groups), quality control (combining defect rates across production lines), and machine learning (deriving mixture distributions).