Form for the exam (Probability and Statistical Inference)

Rule of sum and complement Bravais Correlation Coefficient

P(A∪B) = P(A) + P(B) - P(A∩B) σXY

ρ=
σXσY
P ( A¯ ) = 1 − P ( A )
Independent random variables
Conditional probability
f(x, y) = fX (x)fY(y)
P(A∩B)
P(B|A) =
P(A) Bernoulli random variable, X∼Bern(p)
Product rule f ( x ) = p x (1−p)1−x , x = 0.1
P(A∩B) = P(A)P(B|A) E(X) =p,Var(X) =p(1−p)
P(A∩B) = P(A)P(B) under independence
Random variable Binomial, X ∼ Bin(n, p)
Bayes' theorem n x
f(x) = p (1−p)n−x , x = 0, 1, ..., n
P ( C i )P(A|Ci ) x
P ( C I| A ) = k
j=1P ( C j (P(A|Cj )
n n!
P =
x x!(n−x)!
Expected value of a random variable
E(X) =np,Var(X) =np(1−p)
µ = E(X) = X xf(x) in the discrete
x Poisson random variable, X ∼ Pois(λ)
l2
µ = E(X) = Z xf(x)dx in the continuous λx −λ
f(x) = e , x = 0, 1, 2, . . .
l1 x!

Variance and standard deviation of a random variable E(X) =λ,Var(X) =λ

2
σ 2= Var(X) = X (x−µ) f ( x ) in the discrete Normal random variable, X∼N(µ, σ)2 )
x
1 1 x-µ 2
e− 2 ( σ ) ,
l2
f(x) = √ x∈R
σ 2= Var(X) = Z (x - µ)2 f(x)dx in the continuum σ2π
l1
√ E(X) =µ,Var(X) =σ2
σ=σ2

Standardized random variables Chi-squared random variable, X∼χ2 (r)

X−µ x is in the set of real numbers+
Z= f ( x ) = a ( r ) x r/2−1 e−x/2 ,
σ
E(X) =r,Var(X) = 2r
Marginal probability functions

fX(x) = Random variable Student, X ∼ t(r)

X f(x, y) fY(y) = X f(x, y)
y x
f ( x ) = b ( r ) ( 1 + t 2 /r)−(r+1)/2 , x∈R
Z y)dy,
fX(x) = f(x, Z y)dx
fY(y) = f(x,
E(X) = 0,Var(X) =r/(r−2), r >2
Covariance
Sample media
σXY = X X (x - µX) (y - µY)f(x, y) n
1
x y X̄= Xi
nXi=1
σXY = Z Z (x - µX )(y−µY)f(x, y)dy dx
¯ =µ,Var(X) =σ ¯/n
E(X) 2
Sample variance with known population Confidence interval for perpconX ∼Bern(p),
n great champions
2σ̂ 1
= (Xi−µ)2
nXi=1 (l1 , l2 ) = pˆ∓z1−α/2pˆ(1−pˆ)/n
p

Sample variance with unknown mean A=l2−l1= 2z1−α/2p^(1−p)/n

p
n
1 ¯ 2 z1−α/2 quantile of the standard normal at level 1−α/2
S 2= X (Xi-X) undistorted version
n−1 i=1
n Confidence interval for σ2 conX ~ N(µ, σ2 )
1 ¯ 2
2σ̃
= (Xi-X) distorted version (n−1)s2 (n−1)s2
nX l1= 2
, l2 = 2
i=1 χ1−α/2 χα/2
E(S2 ) =σ2 2 ,χ2
χα/2 1−α/2 chi-square quantiles with n−1 degrees

Sample proportion of freedom at level α/2 and 1−α/2

1
n Hypothesis testing0:µ=µ0 conX∼N(µ, σ2 ),σ2noted
=X
p̂= ¯ Xi ,conXibinary
nX X¯minus mu0
i=1 Statistical test: ZX̄ = ~N(0,1)
p σ /n
2
E(p̂) = p, Var(p̂) = p(1−p)/n
Observed level of significance
Distortion of an estimator T=t(X1 , . . . , X n )
αoss =P(ZX̄ ≥zx̄ |µ=µ0 )= 1−Φ(zx̄) per H1:µ > µ0
D(T) = E(T) - θ, θ unknown parameter
αoss=P(ZX̄ ≤zx̄ |µ=µ0 ) = Φ(zx̄) perH1:µ < µ0
Mean squared error of an estimator T
αossP(|ZX̄ | ≥ |zx̄|) = 2[1−Φ(|zx̄|)] perH1:µ=µ0
MSE(T) = E[(T−θ)2 = Var(T) + [D(T)]2
Hypothesis test H0:µ=µ0 , conX ∼N(µ, σ2 ),σ2no
Int. conf. perµconX ∼N(µ, σ2 ),σ2note note
X¯- µ0
r σ
2 Statistical test: TX̄ =
(l1 , l2 ) =x¯∓z1−α/2 ~t(n−1)
p S /n
2
n
r σ
2
A=l2−l1= 2z1−α/2 Test hypothesis H0:µ=µ0 , with great champions
n
X¯- µ0
z1−α/2 quantile of the standard normal at level 1 − α/2 Statistical test: ZX̄ = ~N(0,1)
p S /n
2

Int. conf. per µconX ∼N(µ, σ)2 ),σ2not noted

Hypothesis testing H0:p=p0conX ∼Bern(p), large
r s
2
(l1 , l2 ) = x̄ ∓ t1−α/2 champions
n
r s −
p̂p0
2 ap
A=l2−l1= 2t1−α/2 Statistical test: Zp̂ = ∼N(0,1)
n p0 (1−p0 )/n
p
t1−α/2 quantile of the Student distribution with n−1 degrees of freedom
Hypothesis test for H0:σ2=σ20 , conX~N(µ, σ2 )
at the level 1−α/2
(n−1)S2
Statistical test: V= ∼χ2 (n−1)
Int. conference for great champions σ20

r s
2
(l1 , l2 ) =x¯∓z1−α/2 Independence test for dimensions×t tables
n
r s
2 Statistical test:
A=l2−l1= 2z1−α/2 n s t 2
(nij-ni0 n0j /n) ap
χ2= XX ∼χ2 ((s−1)(t−1))
z1−α/2 quantile of the standard Normal at level 1−α/2 i=1j=1
ni0 n0j /n
t s
nI0= j1nij n0j = i=1nij
P P