Visual Tools
Calculators
Tables
Mathematical Keyboard
Converters
Other Tools

Probability Methods: Rules, Conditioning, Bayes and Distributions

Every how-to in probability on one page: the basic rules, conditional probability, Bayes, trees, tables and word problems here, and the methods for random variables and distributions gathered from their own hubs. Each card opens its page or its rule card; the rules link back to the lesson where they are explained.

Level
I

Single-rule moves

Atomic · complement, addition, multiplication
1−P

How to find a probability with the complement rule

Complement rule
Atomic
Recognize
The event is easier to describe by what does not happen: "at least one", "not all", "no six".
steps
1. Describe the opposite event AcA^c in words.
2. Find P(Ac)P(A^c) and subtract it from 11.
Statement
P(Ac)=1−P(A)P(A^c) = 1 - P(A)
worked example
At least one head in 33 tosses: 1−P(no heads)=1−(12)3=781 - P(\text{no heads}) = 1 - \left(\frac{1}{2}\right)^3 = \frac{7}{8}
∪

How to find the probability of A or B

Addition rule
Atomic
Recognize
The question says "or": a heart or a king, an even number or a number above 44, rain or wind.
steps
1. Find P(A)P(A), P(B)P(B) and the overlap P(A∩B)P(A \cap B).
2. Add the two and subtract the overlap once; for disjoint events the overlap is 00.
Statement
P(A∪B)=P(A)+P(B)−P(A∩B)P(A \cup B) = P(A) + P(B) - P(A \cap B)
worked example
Heart or king: 1352+452−152=1652=413\frac{13}{52} + \frac{4}{52} - \frac{1}{52} = \frac{16}{52} = \frac{4}{13}
×

How to find the probability of A and B for independent events

Multiplication rule (independent events)
Atomic
Recognize
The question says "and" and one event has no effect on the other: two dice, repeated coin tosses, draws with replacement.
steps
1. Check that the events are independent (separate dice, replacement, separate trials).
2. Multiply the single probabilities.
Statement
P(A∩B)=P(A)⋅P(B)P(A \cap B) = P(A)\cdot P(B)
worked example
Two dice both show a six: 16⋅16=136\frac{1}{6}\cdot\frac{1}{6} = \frac{1}{36}
II

Multi-step methods

Composite · each opens its own page
How to list a sample spaceWrite down every possible outcome of an experiment with a list, a grid or a tree, then count it.uses: outcomes · grid · tree diagramComposite4 steps · 3 examplesHow to find a probability by counting outcomesEqually likely outcomes: count the favorable ones, count all of them, divide.uses: classical probability · complement rule · combinationsComposite4 steps · 3 examplesHow to find a conditional probabilityThe probability of A given B: divide the overlap by the probability of the condition.uses: conditional probability formula · restricted sample spaceComposite4 steps · 3 examplesHow to test whether two events are independentCompare the probability of both with the product of the single probabilities.uses: independence formula · conditional probabilityComposite4 steps · 3 examplesHow to find the probability of A and B for dependent eventsMultiply the first probability by the second one, worked out after the first has happened.uses: multiplication rule · chain rule · conditional probabilityComposite4 steps · 3 examplesHow to use the law of total probabilitySplit into cases, find the chance of the event in each case, weight by the case and add.uses: law of total probability · multiplication ruleComposite4 steps · 3 examplesHow to use Bayes' theoremReverse a conditional probability: from P(evidence given cause) to P(cause given evidence).uses: Bayes' theorem · law of total probabilityComposite4 steps · 3 examplesHow to solve a probability problem with a tree diagramOne level per stage: label the branches, multiply along each path, add the paths you want.uses: tree diagram · multiplication rule · adding disjoint pathsComposite4 steps · 3 examplesHow to find probabilities from a two-way tableJoint, marginal and conditional probabilities: pick the right cell and the right total.uses: joint probability · marginal probability · conditional probabilityComposite4 steps · 3 examples
III

Choosing a rule

Strategy
IV

Word problems

cards, dice, coins, urns
How to solve playing card probability problemsOne card, several cards in a row, or a whole hand from a standard 52-card deck.uses: counting outcomes · multiplication rule · combinationsApplied5 setup steps · 3 examplesHow to solve dice probability problemsSums, doubles and "at least one six": list the ordered outcomes and count.uses: sample space grid · complement rule · conditional probabilityApplied5 setup steps · 3 examplesHow to solve coin toss probability problemsHeads and tails over several tosses: count the sequences, or use the binomial formula for a biased coin.uses: sample space · complement rule · binomial probabilityApplied5 setup steps · 3 examplesHow to solve urn and ball-drawing problemsColored balls drawn from a bag, with or without replacement: multiply step by step or count combinations.uses: multiplication rule · tree diagram · combinationsApplied5 setup steps · 3 examples
V

Random variables

11 pages · 1 rule cards on their own hub
How to find the expected value of a sumThe variable is a sum or a linear combination of variables whose means you know: X+YX + Y, 3X−2Y+53X - 2Y + 5, the total of nn dice. It holds whether or not the variables are independent.uses: Linearity of expectationAtomicrule card on the hubHow to find the expected value of a random variableWeight every value by its probability and add; for a density, integrate xf(x)x f(x) instead.uses: pmf · pdf · weighted averageComposite4 steps · 3 examplesHow to find the variance of a random variableFind the mean, then the mean of the square, and subtract: E[X2]−(E[X])2E[X^2] - (E[X])^2.uses: expected value · computational formula · linear transformComposite4 steps · 3 examplesHow to find the standard deviation of a random variableFind the variance first, then take its square root; the result is in the units of XX.uses: variance · square root · linear transformComposite3 steps · 3 examplesHow to find the covariance and correlation of two random variablesCompute E[XY]−E[X]E[Y]E[XY] - E[X]E[Y] from the joint table, then divide by σXσY\sigma_X \sigma_Y for the correlation.uses: joint distribution · expected value · standard deviationComposite5 steps · 3 examplesHow to find the CDF from a PMFSort the values, keep a running total of the probabilities, and write the step function.uses: pmf · running sum · step functionComposite4 steps · 3 examplesHow to find the CDF from a PDFIntegrate the density from the left end of the support up to xx, and add the flat pieces outside.uses: pdf · definite integral · piecewise functionComposite4 steps · 3 examplesHow to find the median and mode of a distributionThe median is where the CDF reaches one half; the mode is the peak of the pmf or pdf.uses: cdf · pmf · pdf · maximumComposite4 steps · 3 examplesHow to find an expected count with indicator variablesWrite the count as a sum of 0-1 indicators, use E[I]=P(A)E[I] = P(A), and add by linearity.uses: indicator variables · linearity of expectationComposite4 steps · 3 examplesHow to bound a probability with Markov's inequalityFor a nonnegative variable, bound the upper tail by the mean: P(X≥a)≤E[X]/aP(X \geq a) \leq E[X]/a.uses: expected value · tail boundComposite4 steps · 3 examplesHow to bound a probability with Chebyshev's inequalityMeasure the distance from the mean in standard deviations, then bound: P(∣X−μ∣≥kσ)≤1/k2P(|X - \mu| \geq k\sigma) \leq 1/k^2.uses: mean · standard deviation · tail boundComposite4 steps · 3 examplesHow to approximate a probability with the central limit theoremScale the mean and SD to the sum or the average of nn terms, standardize, and read Φ(z)\Phi(z).uses: mean · standard deviation · normal distribution · z-scoreComposite5 steps · 3 examples
VI

Distributions

13 pages · 0 rule cards on their own hub
How to find a binomial probabilityCount successes in nn independent trials: name nn, pp and the values of kk, then add PMF terms or use the complement.uses: binomial PMF · complement rule · binomial CDFComposite5 steps · 3 examplesHow to find a geometric probabilityRepeat a trial until the first success: k−1k - 1 failures then one success, and a short tail formula for "more than kk trials".uses: geometric PMF · geometric CDF · memoryless propertyComposite4 steps · 3 examplesHow to find a negative binomial probabilityThe trial of the rrth success: r−1r - 1 successes somewhere in the first k−1k - 1 trials, then a success on trial kk.uses: negative binomial PMF · binomial coefficient · binomial linkComposite4 steps · 3 examplesHow to find a hypergeometric probabilityDraws without replacement from two groups: choose the successes and the failures separately, divide by all samples.uses: hypergeometric PMF · combinations · complement ruleComposite4 steps · 3 examplesHow to find a Poisson probabilityEvents at an average rate: scale λ\lambda to the interval asked about, then use e−λλk/k!e^{-\lambda}\lambda^k / k! or its complement.uses: Poisson PMF · rate scaling · complement ruleComposite4 steps · 3 examplesHow to find a discrete uniform probabilityEqually likely whole numbers from aa to bb: count the values in the event and divide by b−a+1b - a + 1.uses: discrete uniform PMF · counting integers · discrete uniform CDFComposite4 steps · 3 examplesHow to find a continuous uniform probabilityEqually likely anywhere on [a,b][a, b]: the probability of an interval is its length over b−ab - a, after clipping to the support.uses: continuous uniform PDF · continuous uniform CDF · interval lengthComposite4 steps · 3 examplesHow to find an exponential probabilityWaiting times at a constant rate: get λ\lambda from the mean, then use 1−e−λx1 - e^{-\lambda x} or the tail e−λxe^{-\lambda x}.uses: exponential CDF · exponential mean · memoryless propertyComposite4 steps · 3 examplesHow to find a normal probabilityStandardize with z=(x−μ)/σz = (x - \mu)/\sigma, then read the area from Φ\Phi: left tails, right tails and intervals.uses: standardization · standard normal CDF · symmetry of the bell curveComposite4 steps · 3 examplesHow to find a normal percentileGo backwards from an area to a value: find zz with Φ(z)=p\Phi(z) = p, then x=μ+zσx = \mu + z\sigma; also solve for μ\mu or σ\sigma.uses: inverse standard normal · standardization · quantilesComposite4 steps · 3 examplesHow to approximate a binomial with a PoissonMany trials, small pp: replace Bin⁡(n,p)\operatorname{Bin}(n, p) by Poisson⁡(np)\operatorname{Poisson}(np) and use the short Poisson formula.uses: binomial PMF · Poisson PMF · law of rare eventsComposite4 steps · 3 examplesHow to approximate a binomial with a normalLarge nn: use N(np, np(1−p))N(np,\, np(1-p)), move each boundary by 0.50.5 (continuity correction), then standardize.uses: binomial mean and variance · continuity correction · standardizationComposite5 steps · 3 examplesWhich probability distribution should I use?A decision tree: equally likely values, then counted or measured, then what produces the count or the measurement.uses: leads to 9 distributionsStrategydecision tree