Documente Academic
Documente Profesional
Documente Cultură
DATE STATISTICE
1. Valori calitative;
Exemplu: intrebare cu raspunsuri posibile "f. nemultumit", "nemultumit",
"indiferent", "multumit", "foarte multumit"
n indivizi independenti, alesi in mod aleator dintr-o aceeasi categorie, raspund
la intrebare
> rasp=c("fnem","nem","ind","mul","fmul")
> p=c(0.2,0.3,0.1,0.3,0.1)
> x<-sample(rasp,50,replace=T,prob=p)
>x
"fmul" "ind" "mul" "mul" "nem" "nem" "fmul" "nem" "nem" "nem" "fnem"
"fnem" "nem" "nem" "nem" "mul" "fnem" "fnem" "fnem" "nem" "fnem"
"mul" "fnem" "fnem" "mul" "nem" "nem" "mul" "nem" "mul" "mul" "ind"
"fmul" "mul" "fmul" "fnem" "nem" "nem" "fmul" "nem" "mul" "fnem" "mul"
"nem" "nem" "fnem" "nem" "fnem" "ind" "nem"
2. Valori cantitative
apartinand unei multimi cel mult numarabile de numere reale
apartinand lui R sau unui interval inclus in R
Exemplu: nota obtinuta la un examen ( 0 = absent)
n indivizi independenti, alesi in mod aleator dintr-o aceeasi categorie
> nota=c(0:10)
> p=c(0.05,0,0,0,0.3,0.2,0.15,0.1,0.05,0.1,0.05)
> y<-sample(nota,25,replace=T,prob=p)
>y
4684465597846948444755657
Exemplu: tensiunea arteriala sistolica
n indivizi independenti, alesi in mod aleator dintr-o aceeasi categorie
> z<-c(rnorm(50,13,1.5))
>z
11.4, 14.2, 14.9, 12.5, 12.8, 13.8, 10.7, 13.1, 15.1, 11.4, 11.6, 15.5, 11.8, 12.9,
15.3, 13.7, 13.5, 11.8, 11.9, 12.9, 13.3, 14.2, 14.5, 12.7, 12.4, 13.7, 10.9, 15.4,
14.1, 9.4, 12.5, 11.7, 13.2, 14.9, 14.5, 13.5, 12.5, 13.8, 13.3, 12.8, 10.5, 12.1, 13.5,
14.6, 10.7, 12.1, 10.9, 11.5, 11.7, 11.1
: S ! [0; 1]
P
X
x2A
p (x; ) = 1
x2A
Exemple:
X U f1; :::; rg; r 2 N; r
aruncarea unui zar),
r
X
1
x=1
fxg
X
B (1; ) ; 2 (0; 1) ; A = f0; 1g (ex: aparitia unui "succes" intr-o
proba cu doua rezultate posibile),
P
1
X
1 x
(1
fxg
x=0
X
B (r; ) ; 2 (0; 1) ; A = f0; 1; :::; rg (ex: numarul de "succese" in r
probe independente, cu cate doua rezultate posibile),
P
r
X
Crx
(1
r x
x=0
fxg
X P o ( ) ; 2 (0; 1) ; A = N (ex: numarul de defecte ce pot identicate la piesele dintr-un lot de volum mare),
P
1
X
x=0
x!
exp (
fxg
(fxg) = 0; 8x 2 R
Z
X 1 (B) = f (x; ) dx;
f (x; )
0; 8x 2 R
f (x; ) dx = 1
Exemple:
X
U (0; ) ;
2 (0; 1) ;
1
; x 2 [0; ]
0; x 2
= [0; ]
f (x; ) =
X
Expo( );
2 (0; 1) ;
1
f (x; ) =
X
Gamma ( ; ) ;
2 (0; 1) ;
1
( )
f (x; ; ) =
X
f x; ;
=
2
exp
0;
x 2 [0; 1)
x 2 ( 1; 0)
2 (0; 1) ;
1
exp
x 2 [0; 1)
x 2 ( 1; 0)
0;
;
=p
2R
1
2
(0; 1) ;
exp
densitatea N (0; 1)
1 2
f (x) = p12 exp
2x
1
2
(x
; x2R
0.3
0.2
0.1
0
-5
-2.5
2.5
5
x
F (y) = P
F (y) =
x2A
x<y
(( 1; y)) = P (X < y)
F (y) =
Zy
f (x; ) dx; y 2 R
Exemplu:
X Expo (2)
f (x) =
F (y) =
8
<
:
exp
0;
Ry
0
1
2
exp
x
2
1
2
exp
0;
x
2
y 2 ( 1; 0)
dx;
y 2 [0; 1)
x
2
x 2 [0; 1)
x 2 ( 1; 0)
0;
exp
x
2
y 2 ( 1; 0)
y 2 [0; 1)
0.75
0.5
0.25
0
0
2.5
7.5
10
x
3. Cuantila de rang
a lui X
q )=
8 P r
Z
x p (x; ) ; (< 1) ; pt. rep. discreta
<
Rx2Ar
; r2N
M (X r ) = X r dP =
: x f (x; ) dx; (< 1) ; pt. rep. continua
R
D (X) = M
(X
M (X))
Exemple:
=M
X2
(M (X))
2;
r
X
M (X) =
D2 (X) =
X
B (1; ) ;
1
r+1
=
r
2
x=1
2
12
2 (0; 1) ;
1
X
M (X) =
1 x
(1
x=0
D2 (X) = (1
X
B (r; ) ;
2 (0; 1) ;
M (X) =
r
X
x Crx
r x
(1
=r
x=0
D2 (X) = r (1
X
Po( );
2 (0; 1) ;
M (X) =
1
X
x!
x=0
D2 (X) =
X
U (0; ) ;
exp (
)=
2 (0; 1) ;
Z
M (X) =
dx =
D2 (X) =
X
Expo( );
2 (0; 1) ;
M (X) =
Z1
12
exp
dx =
D2 (X) =
X
Gamma ( ; ) ;
M (X) =
2 (0; 1) ;
1
( )
Z1
2 (0; 1) ;
x x
D2 (X) =
exp
dx =
M (X) = p
D2 (X) =
2R
1
2
Z1
(0; 1) ;
1
x exp
(x
dx =
1
P
p (x; )
x=0
fxg :
GX : [ 1; 1] ! R
GX (t) =
1
X
p (x; ) tx
x=0
(G0X (1))
eitX
M (X) =
ESTIMAREA PARAMETRILOR
Prin alegerea modelului:
forma functionala specicata
existenta unor parametri necunoscuti
"Model parametric"
P
X:
v Rk ; k
; 2
! S; v.a., S = A sau S = R
v.a.i.i.r. P
A ; (P (A)) ;
n
O
R ;B ;
n
O
Xi
Xi
1
i=1
Denitie:
Fie o functie masurabila b : S n
estimator al parametrului :
(X1 ;:::;Xn )
Sn
Xi
i=1
i=1
n
N
Sn; S n;
n
b(x1 ;:::;xn )
Pentru datele statistice (x1 ; :::; xn ) ; valoarea b (x1 ; :::; xn ) se numeste estimatie a lui :
Notatii (presupunand ca toate mediile de mai jos exista):
= ( 1 ; :::;
M
Cov
b; b = cov
= M
0
k)
b = b1 ; :::; bk
b = M
bi ; bj
bi
b1 ; :::; M
bk
i;j=1;:::;k
bi
Pentru. k = 1; M
Denitii:
8
bj
b ; D2 b
bj
i;j=1;:::;k
b (X1 ; :::; Xn ) = ; 8 2
Comentariu:
Pentru k = 1; b (X1 ; :::; Xn ) este ENDM daca
M
b = ; 8 2
D2 b
D2 (g) ; 8 2
= D2 b + Bias b
Denitie:
Fie un sir de observatii i.i.r., (Xn )n si e b (X1 ; :::; Xn )
Spunem ca b este un estimator consistent daca
b (X1 ; :::; Xn ) P!
pentru n ! 1; 8 2
8
>
<
>
:
p (x; )
x2A
fxg ;
caz discret
sau
f (x; ) l; x 2 R; caz continuu
i=1
APLICATIA 1
al repartitiei B (1; )
Modelul
P
1
X
1 x
(1
x=0
Datele statistice
fxg ;
2 (0; 1)
Functia de verosimilitate
L (x1 ; :::; xn ; ) =
n
Y
xi
(1
1 xi
i=1
10
Pn
i=1
xi
(1
Pn
i=1
xi
Constructia EVM
ln L =
n
X
xi ln +
n
X
i=1
1X
@ ln L
=
xi
@
i=1
n
1 X
@ 2 ln L
=
@ 2
xi
i=1
xi
ln (1
n
X
xi
i=1
(1
i=1
)
!
n
X
xi
i=1
@ ln L
=0
@
n
X
b (x1 ; :::; xn ) = 1
xi = x
n i=1
@ 2 ln L
jx =
@ 2
n
<0
x (1 x)
n
X
bV M (X1 ; :::; Xn ) = 1
Xi = X
n i=1
Proprietatile EVM: vom stabili repartitia exacta a estimatorului, vom cerceta nedeplasarea si vom calcula eroarea medie patratica.
Repartitia lui bV M (X1 ; :::; Xn )
Propozitie
Fie variabilele aleatoare independente Yi
Y2 B (r1 + r2 ; )
Rezulta
n bV M (X1 ; :::; Xn ) =
n
X
Xi
B (n; )
i=1
n bV M
=n
D2 n bV M
M
bV M
D2 bV M
= n (1
=
=
(nedeplasare)
(1
n
11
bV M
(1
n
APLICATIA 2
=f (x; ) l
;
0;
x 2 [0; ]
;
in rest
8
< 0;
y
;
FX (y) = P (X < y) =
:
1;
f (x; ) =
M (X) =
2 (0; 1)
y<0
y 2 [0; ]
y>
dx =
D (X) =
x2
dx
12
Datele statistice
L (x1 ; :::; xn ; ) =
L (x1 ; :::; xn ; ) =
xi 2 [0; ] ; i = 1; :::; n
in rest
0;
(
1
n
; 0
0;
maxxi
i
< maxxi
i
Constructia EVM
1
2(0;1)
maxxi
i
se atinge pentru
bV M (x1 ; :::; xn ) = maxxi notat
= x(n)
i
E.V.M. este
12
n
Y
X(n) < y =
F bV M (y) =
f bV M (y) =
i=1
8
<
0;
y n
y<0
y 2 [0; ]
y>
1;
n
n
yn
0;
y 2 [0; ]
in rest
bV M
D2 bV M
M
bV M
yn
dy =
n
n+1
n
n+1
1
n+1
y2
n
n
yn
dy =
n
n+2
n
n+2
n
n+1
Bias bV M
2
2
(n + 2) (n + 1)
2
2
(n + 2) (n + 1)
2
2
(n + 1)
2 2
(n + 1) (n + 2)
n+1
n
D2 b =
b =
n
2
2
(n + 2) (n + 1)
n (n + 2)
n (n + 2)
bV M
b
2n
> 1; n > 1
n+1
<M
13
bV M
2 (0; 1)
b =
=
2
14