Further statistics: formulas
Pick a formula to see it explained with worked examples and a solver that shows every step.
About this topic
Further statistics builds on school statistics with more distributions and tools for combining random numbers. On this page you will find 51 pages, such as Sum of independent random variables, Poisson distribution and Geometric distribution.
You will find clear examples, tips on mistakes and a step-by-step solver on every page. Begin with Sum of independent random variables.
Formulas in this topic
- Expectation and variance of a discrete variableFor a discrete random variable X, E(X)=Σ xp(x) and Var(X)=E(X²)-[E(X)]², where E(X²
- Linear transformation of a random variableFor constants a and b, E(aX+b)=aE(X)+b and Var(aX+b)=a²Var(X).
- Sum of independent random variablesFor any random variables, E(X+Y)=E(X)+E(Y). For independent X and Y, Var(X± Y)=operatorname
- Poisson distributionIf events occur independently at a constant average rate, the number X of events in a fixed interval follows t
- Mean and variance of a Poisson variableFor XsimPo(λ), E(X)=λ and Var(X)=λ. The mean and the variance are e
- Poisson approximation to the binomialWhen n is large and p is small, the binomial distribution B(n,p) is well approximated by Po(np).
- Geometric distributionThe number of trials X up to and including the first success, with success probability p each time, has P(X=r)
- Continuous random variable: pdf and cdfA continuous random variable has a probability density function f(x)≥0 with ∫ f=1. The cumulative distrib
- Mean, variance and median (continuous)For a continuous random variable with density f: E(X)=∫ xf(x)dx, Var(X)=∫ x²f(x)dx
- Continuous uniform distributionA continuous uniform variable on [a,b] has constant density f(x)=dfrac1b-a, mean (a+b)/(2) and varianc
- Exponential distributionThe exponential distribution has density f(x)=λ e^(-λ x) for x≥0, with E(X)=dfrac1λ, ope
- Chi-squared test statisticThe chi-squared statistic compares observed frequencies O with expected frequencies E: chi²=Σdfrac(O-
- Chi-squared: degrees of freedomFor a contingency table with r rows and c columns, nu=(r-1)(c-1). For a goodness-of-fit test with n categorie
- Spearman's rank correlationSpearman's rank correlation coefficient is r_s=1-dfrac6Σ d²n(n²-1), where d is the difference be
- Confidence interval for a mean (σ known)When the population standard deviation σ is known, a confidence interval for the mean is bar x± zdfra
- Probability generating functionThe probability generating function of a discrete variable on the non-negative integers is G(t)=E(t^(X))=Σ
- Combining normal variablesIf Xsim N(μ_X,σ_X²) and Ysim N(μ_Y,σ_Y²) are independent, then aX+bYsim N(aμ_X
- Expected number of outcomesThe expected number of times an event happens is the number of trials times the probability: E=np. Use the cal
- Beta–binomial conjugacyIf θsimBeta(a,b) and Xmidθsim B(n,θ) with X=k, then the posterior is θmid ks
- Inclusion–exclusion (three events)For three events, P(Acup Bcup C)=P(A)+P(B)+P(C)-P(Acap B)-P(Acap C)-P(Bcap C)+P(Acap Bcap C). Add singl
- Law of total probabilityIf A_1,dots,A_k partition the sample space, then P(B)=Σ_iP(Bmid A_i)P(A_i). The probability of B is the
- CovarianceThe covariance of two random variables is Cov(X,Y)=E(XY)-E(X)E(Y)=E[(X-μ_X)(Y-μ_Y)]
- Variance of a sumVar(X+Y)=VarX+VarY+2Cov(X,Y), and more generally \
- Correlation coefficientThe correlation coefficient ρ=dfracCov(X,Y)σ_Xσ_Y measures the strength of th
- Law of total expectation and varianceE(X)=E(E(Xmid Y)) and VarX=E(Var(Xmid Y))+Varb
- Moment generating functionThe moment generating function is M_X(t)=E(e^(tX)). Its derivatives at 0 give the moments: E(Xⁿ)=M_X^((n))(
- MGF of the binomialFor XsimBin(n,p) the moment generating function is M(t)=(q+pe^(t))ⁿ, with q=1-p.
- MGF of the PoissonFor XsimPoisson(λ) the moment generating function is M(t)=exp(λ(e^(t)-1)).
- MGF of the normalFor Xsim N(μ,σ²) the moment generating function is M(t)=exp(μ t+tfrac12σ²t²r
- Chebyshev's inequalityFor any random variable with mean μ and standard deviation σ, P(|X-μ|≥ kσ)≤dfrac1k². A
- Markov's inequalityFor a non-negative random variable X and a>0, P(X≥ a)≤(E(X))/(a).
- Central limit theorem (sampling distribution of the mean)For independent samples of size n with mean μ and standard deviation σ, the sample mean is approximatel
- Weak law of large numbersThe weak law of large numbers says the sample mean converges in probability to the true mean: for every varep
- Gamma distributionThe gamma distribution with shape α and rate λ has density f(x)=dfracλ^(α)x^(α-1)
- Beta distributionThe beta distribution on (0,1) has density f(x)=dfracx^(a-1)(1-x)^(b-1)B(a,b), mean dfrac aa+b and var
- Hypergeometric distributionDrawing n items without replacement from N, of which K are special, the number X of special items drawn has P(
- Multinomial distributionIn n independent trials with k outcomes of probabilities p_1,dots,p_k, the probability of exactly n_1,dots,n
- Change of variable for a densityIf Y=g(X) with g monotonic, then f_Y(y)=f_X(g⁻¹(y))|(d)/(dy)g⁻¹(y)|. Probability
- Joint, marginal, conditional densitiesFor a joint density f(x,y): the marginal is f_X(x)=∫ f(x,y)dy, the conditional is f_Ymid X(ymid x)=d
- Poisson processIn a Poisson process with rate λ, the number of events in time t is Poisson with mean λ t: P(N(t)=
- Markov chain: stationary distributionA stationary distribution of a Markov chain with transition matrix P is a probability vector boldsymbolπ wi
- Markov chain: Chapman–KolmogorovThe n-step transition probabilities are the entries of the matrix power Pⁿ, and P^((n+m))=P^((n))P^((m)): t
- Chi-squared test of independenceFor a contingency table the expected count in cell (i,j) under independence is E_ij=(R_iC_j)/(N) (row t
- Discrete uniform distributionThe discrete uniform distribution on 1,dots,n has P(X=r)=dfrac1n, mean E(X)=dfracn+12 and variance opera
- Expectation of a function of a random variableThe expectation of g(X) is found directly from the distribution of X: E(g(X))=Σ g(x_i)p_i (discrete
- Sum of independent Poisson variablesIf XsimPo(λ_1) and YsimPo(λ_2) are independent, then X+YsimPo(λ_1+l
- Type I and Type II errors, powerA Type I error rejects a true H_0 (probability α); a Type II error accepts H_0 when H_1 is true (probabil
- Quartiles from a cumulative distribution functionFor a continuous random variable with distribution function F, the quartiles solve F(Q_1)=0.25, F(m)=0.5 (the
- Yates’ continuity correction (2 × 2 table)For a 2 × 2 contingency table, Yates' correction replaces each term by dfrac(|O-E|-tfrac12)²E. It red
- Exponential distribution: cumulative distribution functionFor an exponential distribution with rate λ: F(x)=1-e^(-λ x), P(X>x)=e^(-λ x), and the distr
- Exponential waiting times in a Poisson processIf events occur as a Poisson process of rate λ, the gaps between successive events are independent expon
Further statistics questions
What is an expected value?
It is the average result you would get if you repeated something many, many times.
What if I get stuck?
Read the worked examples on the page first, then try the solver with your own numbers and compare each step.