Formulari d’Estadı́stica

∑
n ∑
n
1∑ 2
n
x= 1
n
xi s2X = 1
n
(xi − x) = xi − x2
2

i=1 i=1
n i=1
∑
p ∑n
x= xi f i on fi = fr (xi ) n
ŝ2X = n−1 1
s2X = n−1 x2i − n−1
n
x2
i=1 i=1

∑
n
sX Y = 1
n
xi yi − x y
i=1
∑
p
∑
q sX
Coef. Variació =
sX Y = xi yj fij − x y |x|
i=1 j=1
on fij = fr (xi , yj )

k ∈ (0, 100), n = k
(N − 1) + 1
k ∈ (0, 100), n = k
100
(N − 1) + 1 100

Pk = li o
Pk = x n o
li+1 − li
Pk = xi + (xi+1 − xi )(n − Ni ) Pk = li + (n − Ni )
Ni+1 − Ni

Si X v.a. discreta, Si X v.a. contı́nua,

P [X = x] = f (x) ∑ P [X = x] = 0
∫ t
F (t) = P [X ≤ t] = f (x) F (t) = P [X ≤ t] = f (x) dx
x≤t −∞

Si x = h(y), σX2 = var(X ) = E [(X − E[X ])2 ]

fY (y) = fX (h(y)) · |h′ (y)| = E[X 2 ] − (E[X ])2

∑ ∑
µX = E[X ] = xf (x) E[h(X )] = h(x)f (x)
∫ x ∫x
= xf (x) dx = h(x)f (x) dx
IR IR

E[Z] = n · E[X ] E[X ] = E[X ]

∑
n
1∑
n
Z= Xi , √ X = Xi , σX
n i=1 σX = √
i=1 σZ = n · σX
n
 
 0 si t < 0
Bernoulli, X ∼ B(p) E[X ] =
√p
F (t) = 1 − p si 0 ≤ t < 1
f (x) = px (1 − p)1−x , x = 0, 1  σX = p(1 − p)
1 si t ≥ 1


 0 si t < 0

(n)Xx ∼ B(n,n−x
Binomial, p)  ∑[t]
E[X ] =
f (x) = x p (1 − p) F (t) = f (x) si 0 ≤ t ≤ n √np

 σX = np(1 − p)
per a x = 0, 1, . . . , n 
 x=0
1 si t > n
{
Geomètrica, X ∼ G(p) 0 si t < 1 E[X ] =
√1/p
F (t) =
f (x) = p(1 − p)x−1 , x ∈ IN ∗ 1 − (1 − p)[t]
si t ≥ 1 σX = (1 − p)/p

Poisson, X ∼ P (λ) 
 0 si t < 0
∑ λx
[t] E[X ] =
√λ
λx F (t) =
f (x) = e−λ , x ∈ IN 
 e
−λ
si t ≥ 0 σX = λ
x! x=0
x!

Uniforme discreta a {a1 < · · · < an } 

0 si t < a1
1∑
n
X ∼ U {a1 , . . . , an } ♯{ai ≤ t}
F (t) = si a1 ≤ t ≤ an E[X ] = ai
1 
 n n i=1
f (x) = , x ∈ {a1 , . . . , an } 1 si t > an
n


Uniforme contı́nua, X ∼ U [a, b] 
 0t − a si t < a a+b
E[X ] =
1 F (t) = si a ≤ t ≤ b 2 √
f (x) = , x ∈ [a, b] 
 b−a σX = (b − a)/ 12
b−a 1 si t > b
{
Exponencial, X ∼ E(λ) 0 si t ≤ 0 E[X ] = 1/λ
F (t) =
f (x) = λe−λx , x > 0 1−e−λt
si t > 0 σX = 1/λ

Normal, X ∼ N (µ, σ) ∫ t
(x − µ)2 F (t) = f (x) dx
E[X ] = µ
1 −
f (x) = √ e 2σ 2 , x ∈ IR −∞ σX = σ
σ 2π

Gamma, X ∼ G(α, λ) ∫ t
λα α−1 −λx E[X ] =
f (x) = x e , x>0 F (t) = f (x) dx √ α/λ
Γ(α) 0 σX = α/λ
amb Γ(α) = (α − 1)! si α ∈ IN ∗

xi-quadrat, X ∼ χ2n F de Fisher, X ∼ Fn1 ,n2 t de Student, X ∼ tn

√ √ √
n2 2n22 (n1 +n2 −2) n
E[X ] = n, σX = 2n E[X ] = n2 −2
, σX = n1 (n2 −2)2 (n2 −4)
E[X ] = 0, σX = n−2
 S λa S
Per a qualsevol sistema: T = 1/λ S = 1/µ σ= ρ= pn = P[N = n]
T c

X =Q+S N = λa X Nq = λa Q Ns = λa S N = Nq + Ns

FQ (t) = P[Q ≤ t] FX (t) = P[X ≤ t]

Taula 1. Fórmules per a sistemes de cues M/M/1

λ
σ=ρ= = λS λa = λ pn = (1 − ρ)ρn P[N ≥ n] = ρn n≥0
µ

S ρS ρ ρ2
X= Q = ρX = N = λX = Nq = λQ =
1−ρ 1−ρ 1−ρ 1−ρ

FQ (t) = 1 − ρe−(µ−λ)t = 1 − ρe−t/X FX (t) = 1 − e−(µ−λ)t = 1 − e−t/X t≥0

Taula 2. Fórmules per a sistemes de cues M/M/1/K

λ
σ= = λS
µ

 (1 − σ)σ n

 1 − σ K+1 si λ ̸= µ
pn = per a n = 0, 1, . . . , K, i pn = 0 per a tot n > K


 1 si λ = µ
K +1

pn
p∗n = n = 0, . . . , K − 1 λa = (1 − pK )λ ρ = λa S = (1 − pK )σ
1 − pK

 1 − (K + 1)σ K + Kσ K+1

 σ si λ ̸= µ
(1 − σ)(1 − σ K+1 )
N= Nq = N − (1 − p0 )
K

 si λ = µ
2

N Nq
X= Q=
λa λa
[ ] [ ]
∑
K−2 ∑
n
(µt)i ∑
K−1 ∑
n
(µt)i
FQ (t) = 1 − e−µt p∗n+1 FX (t) = 1 − e−µt p∗n t≥0
n=0 i=0
i! n=0 i=0
i!
 Taula 3. Fórmules per a sistemes de cues M/M/c

λ
σ= = λS ρ = σ/c λa = λ
µ
 n
[ c−1 ]−1  σ

 n! p0 si n = 0, 1, . . . , c − 1,
∑ σn σc
p0 = + pn =
n! c!(1 − ρ) 
 n
n=0  σ p0 si n ≥ c
c!cn−c

σc
c! σc
C(c, σ) = P[N ≥ c] = = p0 (fórmula C d’Erlang)
∑ σn σc
c−1 c!(1 − ρ)
(1 − ρ) +
 [ c−1 n=0 n! ] c!

 ∑ σk σc
 p0 + si 0 < n < c,
P[N ≥ n] = k! c!(1 − ρ)


k=n

C(c, σ)ρn−c si n ≥ c

∑
∞
ρC(c, σ)
N = λX = Nq + σ Nq = (n − c)pn =
n=c
1−ρ

Nq C(c, σ)S
Q= =
λ c(1 − ρ)

FQ (t) = 1 − C(c, σ)e−µt(c−σ) t≥0


 c − σ − 1 + C(c, σ) −µt C(c, σ) −µt(c−σ)
1 − e + e si σ ̸= c − 1 t ≥ 0,
FX (t) = c−σ−1 c−σ−1


1 − [1 + C(c, σ)µt]e−µt si σ = c − 1 t ≥ 0

Taula 4. Fórmules per a sistemes de cues M/M/∞

λ σ n −σ
σ= = λS λa = λ pn = e n = 0, 1, 2, . . .
µ n!

N =σ Q = Nq = 0 X=S FX (t) = FS (t)

Totes les fórmules anteriors són també vàlides per als sistemes de cues M/G/∞
 Taula 5. Fórmules per a sistemes de cues M/M/c/c (o M/M/c amb pèrdues)

σn
λ n!
σ= = λS pn = n = 0, 1, . . . , c, i pn = 0 per a tot n > c
µ σ2 σc
1+σ+ + ··· +
2! c!

σc
B(c, σ) = P[N = c] = c! (fórmula B d’Erlang)
σ2 σc
1+σ+ + ··· +
2! c!

pn λa S
p∗n = n = 0, . . . , c − 1, λa = λ (1 − B(c, σ)) ρ=
1 − pc c
N
N = λa S = σ[1 − B(c, σ)] X= =S Q = Nq = 0 FX (t) = FS (t)
λa

Excepte l’última, totes les fórmules anteriors són també vàlides per als sistemes de cues
M/G/c/c

Taula 6. Fórmules per a sistemes de cues M/M/c/K

λ
σ= = λS r = σ/c
µ
 n
 σ

 n! p0 si n = 0, 1, . . . , c
pn = i pn = 0 per a tot n > K

 c
 σ rn−c p0 si n = c + 1, . . . , K
c!
[ ]−1

 ∑c
σ n
σ c
r(1 − r K−c
)
[ ]−1 
 si r ̸= 1


+
1−r
∑
c
σn σ ∑
c K−c
n=0
n! c!
n
p0 = + r = [ ]−1
n! c! 
 ∑c
n=0 n=1 
 σ n
σ c

 + (K − c) si r = 1
n=0
n! c!

pn λa S
p∗n = n = 0, . . . , K − 1 λa = λ(1 − pK ) ρ= = (1 − pK )r
1 − pK c

σ c rp0 [ ]
N = Nq + λa S Nq = 1 + (K − c)r K−c+1
− (K − c + 1)r K−c
c!(1 − r)2

N Nq
X= Q=
λa λa
 Taula 7. Fórmules per a sistemes de cues M/M/1//m (o M/M/1/m/m)

O = 1/ω S = 1/µ ζ = µ/ω

[ ]−1
∑
m
m!
p0 = ζ −n = B(m, ζ), sent B(m, ζ) la fórmula B d’Erlang
n=0
(m − n)!



 ζ m−n



 m! (m − n)!

 ζ −n p0 = m k si n = 0, 1, . . . , m
(m − n)! ∑ζ
pn =

 k!

 k=0



0 si n > m

ρ = 1 − p0 = 1 − B(m, ζ) λa = µρ = ω(m − N )

m 1
X= − Q=X −S N = λa X Nq = λa Q
λa ω

ζ m−n−1
(m − n) (m − n − 1)!
p∗n = pn = n = 0, . . . , m − 1
m−N ∑ ζk
m−1

k=0
k!

Q(m − 1; ζ + µt) Q(m − 2; ζ + µt)

FX (t) = 1 − FQ (t) = 1 − t≥0
Q(m − 1; ζ) Q(m − 1; ζ)
∑n
en què Q(n; y) = e−y k=0 y k /k! = 1−FY (y), essent Y una VA gamma de paràmetres α = n+1
iβ=1

Taula 8. Fórmules per a sistemes de cues M/G/1

λS 2 λ2 S 2
λa = λ ρ = λS Q= Nq = λQ N = λX = +ρ
2(1 − ρ) 2(1 − ρ)

Taula 9. Fórmules per a optimització de cues

√
λ Cclient
cues M/M/1 µ=λ+ optimitzar µ amb λ fix
Cµ
Cservidor
cues M/M/c N (c) − N (c + 1) ≤ ≤ N (c − 1) − N (c) optimitzar c amb λ i µ fix
Cclient
 Error Tipus I Error Tipus II

e1 = P [rebutjar H0 /H0 certa] = PH0 [Wα ] e2 = P [acceptar H0 /H1 certa] = PH1 [W α ]

∑
k
(oi − ei )2 ∑
k
o2
Test χ 2
e i = nT · pi EPχ2 = χ20 = = i
− nT
i=1
ei i=1
ei

Kolmogorov - Smirnov EPKS = D0 = max{|Fk (x) − F̂ (x)|, |Fk−1 (x) − F̂ (x)|}

∑
n
sX Y s2e = 1
(yi − ŷi )2
y − y = 2 (x − x) n−2
sX
sX Y (i=1n )
y = bx + a on b = 2 i a = y − bx ∑ ∑
n ∑
n

sX = 1
n−2
yi2 − a yi − b xi y i
i=1 i=1 i=1

√
1 (x0 − x)2
sX Y ŷ0 ± tα/2 se + +δ
n ns2x
rX Y = {
sX · sY 0, si és per a E[Y0 ]
on δ =
1, si és per a y0