13 Variable aléatoire uniforme
Oliver Sonnentag, PhD: GÉO1512 – Géographie Quantitative I Séance 4: 22 octobre 2012
f(x)=
1
a−b when a ≤ x ≤ b
0 when x < a or x > b
"
#
$
%
$
• avec a = limite inférieure
• avec b = limite supérieure
http://www.r-tutor.com/
14 PDF et CDF
Oliver Sonnentag, PhD: GÉO1512 – Géographie Quantitative I Séance 4: 22 octobre 2012
• f(x) = fonction de distribution de probabilité
(PDF): PDF d'une variable aléatoire
continue est l'attribution de probabilités
qu'une variable aléatoire continue X se
produit dans un intervalle I [a, b]
• F(x) = fonction de distribution cumulative
(CDF): CDF d’une variable continue X
est F(x) = P(X < x)
!
• PDF is the derivative (i.e., rate of change)
of the CDF.
Non-negative
integrable
g
pdf
f
Normalize by integral of g
CDF
F
Integrate Differentiate
and one place that’s true is when it comes to defining expectations. Remember
that for discrete variables
E[X]⌘X
x
xp(x)
For a continuous variable, we just substitute f(x) for p(x) and an integral for a
sum:
E[X]⌘Z1
1
xf(x)dx
All of the rules which we learned for discrete expectations still hold for contin-
uous expectations.
Let’s see how this works for the uniform-over-[0,10] example.
E[X]=Z1
1
xf(x)dx =Z10
0
x1
10dx =1
10
1
2⇥x2⇤10
0=1
10
1
2(100 0) = 5
Notice that 5 is the mid-point of the interval [0,10]. Suppose we had a uniform
distribution over another interval, say (to be imaginative) [a, b]. What would the
expectation be? First, find the CDF F(x), from the same kind of reasoning we
used on the interval [0,10]: the probability of an interval is its length, divided by
the total length. Then, find the pdf, f(x)=dF/dx; finally, get the expectation,
11
Cumulative probabilities provide, for each value x, the probability of a result less
than or equal to X
15 Variable aléatoire normale (Gaussian)
Oliver Sonnentag, PhD: GÉO1512 – Géographie Quantitative I Séance 4: 22 octobre 2012
• La distribution de probabilité la plus familière
• Constitue le fondement théorique de la régression linéaire et analyse
de la variance (ANOVA) ! séances 9 & 10
• Defined by two parameters (µ, σ):
! E(X) = µ ! central tendency
! σ2(X) = σ2 ! spread around the central tendency
• Variable aléatoire normale (“variable aléatoire de Gauss”): X ~ N(µ, σ)
• Standard normal distribution: µ = 0 et σ = 1
! Variable aléatoire normale standard (Z): E(Z) = 0, σ2 = 1