11 Random experiment Probability Counting → ↳ consider multiple Product rule If : Generalization ⇒ Permutation • all N stages ! ✗ hz ! ✗ Combination ↳ distinct to order . _ . )! ✗ group A# is n PCE / F) = partition r items Probability = sample space to:P(A) =P (AIB ) PCB) t Bayes Independent Events : ↳ Mutually exclusive Formula ↳ and M n , . total objects where n n , one a # = 0 {Fi alike , Nz . . . selection does not matter . (1) = of group that r given F occurred PCE / F) , , Fz , . . . , PCE I F) PCE F) = Az) PCA 1) P(Az Ai ) Fn } is PCE / F) , mutually exclusive PCFIEJPCEJ = , , = = → 1 if group of F becomes r the that does not . " sample space = , , and F if one occur = , . . . PCE IF;) PCF;) 27--1 PCE / Fi ) PC Fi) PCE F) PCE ) PCF) . = = . , . , = . At AzAs) ≤ ¥ , PCEI Fit PEE it = " FCE PCAAZA } ) =P (A) PLAY A) PCA} then PCE) P (Fj / E) I PC F) Event E ☒ if ECF = t # × contains PpÉ¥ = PCA : : × object as PfAli:3 4. PCB 4 PCE / F) : that the order of such group independent cannot together / independent events are unaffected If Pl Ai) =p PCAa) =p ,PCAn)=pn ⇒ P (A) 1T¥ , ( pi ) k trials (E) pka-p.tn e.g n in dependant trials success prob =p K successful . . ! - ↳ m n = ! particular objet label : of E- PCEF) 0 if of . + = argument of Cr n Sequential Conditioning formula if ☒re T.li?=ilAkl--1AilX1AzIx...xIAvl : ↳ Conditional . (E) (4) ' # = K members from ! r Pascal 's Identity :(F) Combinatorial ↳ realizes ✗ Nr ! Selecting : r items the outcomes Number of different permutations of : = - n stage important one n ! Cn experiment ! n Ri outcome before random experiments stages of which has two where outcome of Kth stages Multinomial coefficient ↳ specific Joint and the distinct items identifying unknown before realization i possible outcomes experiment ways : outcomes Quantifies chance of → enumerating - multiple possible by others L2 sample space ↳ notations Evert ↳ : set of all s : subset of : Empty event { = 42,3 } E is , , (O ) null event contained Properties : ( 0 d) , in F M if EF EU F : law Associative law De Principle of Some experiment = Hor T , j = H or T } , { all ordering of A , B , C } Morgan 's 0 ≤ : . PCE ) ' In Ex . sample space : = ETF , ∅ E and F , Ed F law mutually EF (EX F) one in F Viki E- i , E and F . exclusive are . / Min # Ei Ec , identical if ECF , E- of - - P (E) 1- = are - ≤1 PCE ) P(UH , Ei ) has = = - Generalized Form : i : - Distributive law Properties 's ) , ECF if all outcomes ME : Commutative Axioms of Prob {( i an sample space = Union and Intersection ↳ outcomes of possible equally (U ¥ , E) Eni PCEI) if PCEU F) =P(E) 1- PCF) PLEA F) , PCs ) =L P , = = , all are mutually exclusive - , = likely outcomes : P {Ei } = P{Ez} = - - =p {En } = YN Good luck 14 Random variable Discrete Random : var possible whose . Probability Mass Function values Ki } : ." i ki = - value , a> : = Y ↳ ECXY] Eij (Kisi) (pi Ei ) = Variance Properties Var [× ] Var (b) : X and Y ↳ if Bernoulli P(✗ = K) P ( ✗= K ) ↳ EH] G. RV is K - - (ni p) → = xi I "' - P) Yp p E [✗ ] 2 , memory less : = trials n , . . if F- ✗ . (Elly) 2 ( 2- p ) / p2 , ×> Vorcax) , NOT p (1) = 0.2 E [✗ it . . . . value . 13=12×0.2 ) TAKE [✗ it b for ALL X , Y C- [×] = = of Var ( X ) P{ ✗=L} =p P {✗ 03=1 p Then = - , var (X) =p =p , - p . 2 experiment . , first E- [×] , = =P( ✗ = up independent one success Vor (X ) ) expected then , umps pti) Vor (X ) 1- Vor ( Y ) Cnz P ) n µ , E= , P(X=k) . a, = Ki ) and I E [ Y] . k successes get , and 2 . , . - - = a- ) p . P(✗ ntkl = infinite but countable seq intervals [Kit on AKXK tb ] ✗=L , failure is 11=0 1<=1 , 2,3 , an ) P(Y=yi ) EH] Vor (Xt 4) Number of trials needed to ( = and Xzn Binomial , Ii ) - number of successions of Bernoulli (E) pka Hn = = t - - E [X2] = In .pt On : = : , success is n If XY Binomial Geometric Y=yi ) =P (✗ - var (✗ 1- b) = Vor CX) : - , - ↳ ECXN] of Bernoulli Binomial xi a P( ✗ ≤a. Y ≤ b) =P ( ✗ ≤ a) P( Yes ) ⇒ (Ei Ri pi ) (Ejyjoej ) , trial a constant is finite sequence i) weighted average of all possible values : E [( X µ )2] = 0 F , " ≤• independent independent one suppose : = = = - : are PC ✗ ⇒ u 27=1 Pki) =L ECX] tb , E [A. ✗ it a . ↳ discrete ) → as E=ii (e.g p (ki ) Linearity E [ ctb ] Independence Ru X and : mean , Pki ) = Expectation ( exerted _ be written can ki =P {✗= p( ) Cummilatine Distribution Function : f- (a) Lf E[✗] real numbers Mapping sample space to : denoted → then by = npctp) ✗ it ✗ an B Cnitnz , P ) ✗ I = ( 1= , Vor ( X ) , ) K p ) / p2 ⇒ still need K trials n after fails . Notes