Download as pdf
Download as pdf
You are on page 1of 6
Table of Common Distributions Discrete Distributions Bernoulli(p) pmf P(X =2\p)=p"(1—p)!%; z= vd inne EX =p, VarX =p(l—p) maf Mx(t) = (=p) + pe! Binomial(n,p) pmf P(X =aln,p) = (2)p"(1—p)*;, 7 =0,1,2,....0) OSpS1 ‘mean and tariance, EX = mp, VarX = np(1—p) maf Mx(€) = [pe +0 )]” notes Related to Binomial Theorem (Theorem 3.2.2). The multinomial distri- bution (Definition 4.6.2) is a multivariate version of the binomial distri- bution. Discrete uniform pmf P(X=a|N)=}; 2=1,2,...,N; N=1,2,. mean and yyy yar gp x — Wa) variance a 1 maf Mx() = OE Geometric(p) pmf P(X =alp) =p(l— py; 2 =1,2,...; OSps1 ‘mean and 1 14 mariance EX = 5 VarX = {5h mof Mx(t) = tga t < —log(1-p) notes Y = X —1is negative binomial(1,p). The distribution is memoryless: P(X > s|X >t) =P(X > 8-2) Hypergeometrie pmf P(X = a|N,M,K) = Oe 2 =0,1,2,...,K5 M-(N-K)0 meanand yy _ KM = KML == Ke) variance EX = "ys VarX = “3° So=ay notes If K & M and N, the range = 0,1,2,..., K will be appropriate. Negative binomial(r,p) pin (2) =p); r=0,1,...5 OSpsl mean and py _ r0-) -yap x = =P) variance P P mof Mx(t) = (=e) ; t<—log(1~p) notes An alternate form of the pmf is given by P(Y = ylr,p) = (¥})p"( vp)", y =rr-+1,.... The random variable ¥ = X +r. The negative binomial can be derived as a gamma mixture of Poissons. (See Exer- cise 4.32.) Poisson(>) pmf P(X =a)d)=22¥; 2 =0,1...5 0S A<00 mean and py), VarX = variance mof Mx(t) = eM) ‘TABLE OF COMMON DISTRIBUTIONS 623 Continuous Distributions Beta(a, B) pdf f (cla, 8) = wey22 (1-281, O0, B>0 mean and a variance X= ate VerX = apiarery maf Mx() =1 4 DE, (TES 8) & notes ‘The constant in the beta pdf can be defined in terms of gamma functions, Bla, 8) = KOPE). Equation (3.2.18) gives a general expression for the moments. Gamma(a, 8) pdf F(tla,B) = pages te !?, OSa0 mean and py _ ag? nein ond BX = as, VarX = af maf Mx(= (sn). <4 notes Some special cases are exponential (a = 1) and chi squared (a = p/2, B = 2). fa = 3, ¥ = V/X/B is Mazwell. ¥ = 1/X has the inverted gamma distribution. Can also be related to the Poisson (Example 3.2.1). Normal(1:,07) odf J (aly, 07) = Gece @W 0), 00 <2 <0, -00

0 mean and By VarX = variance mof Mx (t) = entto?@/2 notes Sometimes called the Gaussian distribution. Chi squared(p) pdf (el) = reppae?Ne7?; OS a <0; p=1,2,.. mean and lar X = varianee, EX =p, VarX = 2p pit maf Mx()= (en). t<4 notes Special case of the gamma distribution. Uniform(a,b) pdf (ela, 6) mean and py _ bia, Var X = Goa? variance mof Mx(t) = notes ‘If a= and b= 1, this is a special case of the beta (a = 9 = 1) Double exponential(u,0) pdf S(elu,o) = zee, 00 0 meanand py 4 Var X = 202 variance mof Mx(t) = ap, (<3 notes Also known as the Laplace distribution. Exponential(3) pdf F(al8) = Ze*/®, OS a< co, B>0 mean and py og VarX = variance mgf Mx(t)= shy, t<4 notes Special case of the gamma distribution. Has the memoryless property. Has many special cases: Y = X1/7 is Weibull, ¥ = /2X/G is Rayleigh, Y =a ~ ylog(X/A) is Gumbel. Cauchy(0,0) pdf Melo) = Spday cos Ec; 00 0 mean and do not exist variance mof does not exist notes Special case of Student's t, when degrees of freedom = 1. Also, if X and Y are independent n(0,1), X/Y is Cauchy. F ers wan zl +Y) = FETs aE rif Halas) = ES wan O4 moments ARP) yyy » (maf does not exist) PX" = aati rz) (a)") 2<# notes Related to chi squared (Fiyyy = (2) / (2). where the y2s are in- dependent) and ¢ (F,,, = #2). t mS) ea mina eran oot, vex (maf dacs not ent) PX" = SERGE D 0! itm van ven EX" =Oif'n

0, B>0 mean and variance, EX= Bot 8>1, VarX=goiigesy, >2 maf does not exist Logistic()1, 8) af Stl.) = bpp, 00 <4 < 00, —c00 mean and py oy VarX = ne variance mof Mx(t) = eT(1— BT (A+ Bt), |<} notes ‘The cdf is given by F(x|u,8) = —o=t=ay0- Lognormal(s.,02) pdf S(alu,07) ye ee, O<20 mean and py — eut(o*/2), Yar X = etleto® variance moments m= enutnte? /2 (maf does not exist) PX" =e" notes Example 2.3.5 gives another distribution with the same moments. Weibull(,) paf S(aly,B) = Fat te-P/8, OSa0, B>0 mean and BX = gor ( +2), Varx= 67 Ir ( +2)-r (14+ 1) moments EX" = S/T (Q + 2) notes ‘The mgf exists only for 7 > 1. Its form is not very useful. A special case is exponential (7 = 1).

You might also like