2rvariables Slides

Outline
1. Definition of random variable

2. Discrete and continuous random variables
Discrete random variables (probability mass function and cdf)
Continuos random variables (density mass function and cdf)
Random variables
3. Characteristic features of a random variable

Location, scatter and shape parameters
4. Transformations of random variables

5. Random vectors
Joint distribution and independence
Characteristic features, covariance matrix
Ignacio Cascos
Depto. Estadstica, Universidad Carlos III
Definition of random variable
Ignacio Cascos
A random variable assesses a real number to

each possible outcome of the random experiment.
Definition. A random variable X is a mapping

X: E IR, where E is the sample space
associated with a random experiment.
X
e1
It is random because we do not know its value

before carrying out the random experiment.
e2
e3
X (e3)
Ignacio Cascos
Ignacio Cascos
X (e2)
X (e1)
IR
4
Outline
The events are now of the type XA,

where A is a subset of IR.

Their probabilities are P(XA) = P({eE: X(e)A}).

Properties:
1.
P(XA) 0 ;
2.
P(XIR) = 1 ;
3.


5. Random vectors
if A1, A2,IR satisfy AiAj= for i j,

then P(Xi=1, Ai)=i=1, P(XAi) .

Ignacio Cascos
Discrete and continuous random variables
Examples: number of defective items, number of rolls of a

die until a 5 occurs.
A random variable is continuous if its range contains

an interval.
Example: battery life.
Discrete random variables

Given a discrete random variable X, its probability mass
function assigns to each possible value of the variable
the probability that X assumes such a value.
p: IR [0,1]
The range of a random variable is the set of values that

it can assume.
A random variable is discrete if its range is finite or
denumerable.
Ignacio Cascos
Ignacio Cascos
x p(x) = P(X=x)
It satisfies that 0 p(x) 1 for every x and if X assumes
n different possible values x1,,xn , then i p(xi) = 1.
We have P(XA) = xiA p(xi) .
Ignacio Cascos
Let X be the number of heads

after tossing 3 times a fair
coins. The probability mass
function of X is:
The cumulative distribution function (cdf) of X

at x is the probability that X is smaller or equal
to x.
F(x) = P(X x)
0.375
0.05
0.20
p(x) = P(X=x)
0.125
0.375
0.125
0.15
x
0
0.00
probabilidad
0.25
0.30
0.35
Probability
function
Funcion demass
probabilidad
0.10
2.
3.
0
Ignacio Cascos
limx F(x) = 0 ;
limx F(x) = 1 ;
F is nondecreasing ;
F is right-continuous.
1.
4.
numero motores averiados
Ignacio Cascos
10
Continuous random variables
The cumulative
distribution function of
a discrete random
variable is stepwise,
Since the range of a continuous random variable is not

denumerable, an expression like i p(xi) = 1 makes no sense.
0.8
1.0
Funcion de distribucion
cdf
0.6
Histogram for the lifetime of 10000 batteries.

Histogram of duracion
0.2
0.2
numero motores averiados
0.0
0.0
-1
Ignacio Cascos
0.4
Density
0.4
Density
0.0
F(x) = P(X x)
= xi x p(xi)
0.2
0.6
0.6
0.8
0.8
Histogram of duracion
0.4
probability
probabilidad
11
Ignacio Cascos
4
duracion
4
duracion
12
Function f describes the

curve drawn together
with the histogram on
the right. We have
P(2 X 4) 24 f(x)dx
where X is the lifetime of
a battery.
The density mass function f describes the probability

distribution of a continuous random variable. It satisfies:
Ignacio Cascos
0.8
1.0
0.6
1.
0.4
2.
0.2
3.
f(x) 0 ;
+ f(x)dx = 1 .
We have P(a X b) = ab f(x)dx .
Given X continuous r.v., it satisfies
0.0
density
densidad
P(X = a) = 0 ;
P(a X b) = P(a < X b) = P(a X < b) = P(a < X < b)
13
Ignacio Cascos
14
In order to compute the cumulative distribution

Function (cdf) of a continuous random variable,
we must integrate its density mass function,
F(x) = P(X x) = x f(t)dt
Since the cumulative distribution

function is a primitive of the
density mass function, deriving
the cdf, we obtain the density
mass function,
f(x) = dF(x)/dx .
2.
3.
4.
limx F(x) = 0 ;
limx F(x) = 1 ;
F is nondecreasing ;
F is continuous .
Ignacio Cascos
We are working with

f(x) = ex if x > 0
F(x) = 1 ex if x > 0
15
Ignacio Cascos
cumulative probability
1.
Exponential Distribution
1
Mean
1
0,8
0,6
0,4
0,2
0
-1 0
16
Outline
Mathematical expectation or mean

The expecation or mean () of a random variable

is a central point with respect to its distribution

X discrete, = E[X] = xip(xi)
X continuous, = E[X] = xf(x)dx

Properties: Given X,Y and two real numbers, a,b
1. E[a+bX] = a+bE[X] ;
2. E[X+Y] = E[X]+E[Y] .

5. Random vectors
Ignacio Cascos
17
Ignacio Cascos
18
Mathematical expecation or mean
Median
Given a function g: IR IR, the expectation of

random variable g(X) can be computed as
The median of a random variable X is a value Me

such that
F(Me) 1/2 ; P(X Me) 1/2
X discrete, E[g(X)] = g(xi)p(xi)

X continuous, E[g(X)] = g(x)f(x)dx
Ignacio Cascos
If X is a continuous r.v., then F(Me) = 1/2.
19
Ignacio Cascos
20
Quantiles
Scatter parameters
For 0 < < 1, the -quantile of random variable X is a

value x such that the probability of X being not greater
than x is, at least, and the probability of X being not
smaller than x is, at least, 1.
The variance of a random variable X is given by

2 = Var[X] = E[(XE[X])2]
F(x) = P(X x) ; P(X x) 1

Percentiles and quartiles are defined in a similar way
Pa = xa/100 ; Qi = P25i
21
Scatter parameters
22
Describe features from the distribution of a r.v.

different from location and scatter
Given a,bIR and a random variable X, we have

the following properties of the variance
1. Var[b] = 0 ;
2. Var[aX] = a2Var[X] ;
3. Var[aX+b] = a2Var[X] .
Ignacio Cascos
Shape parameters
Property. Var[X] = E[X2]E[X]2 = E[X2]2
Ignacio Cascos
X continuous, 2 = Var[X] = (x)2 f(x)dx

The standard deviation is the (positive) square
root of the variance, = (Var[X])1/2 .
where1 a 99 and 1 i 3 .
Ignacio Cascos
X discrete, 2 = Var[X] = (xi)2 p(xi)
23
k-th moment about the origin, mk = E[Xk]

k-th moment about the mean, k = E[(X)k]
Skewness. Skew = 3/3

Kurtosis. Kurt = 4/43
Ignacio Cascos
24
Example (Flange thickness)
Outline
The thickness of a flange on an aircraft component, X in mm, has

the following density mass function

200( x 0.95) if 0.95 < x < 1.05

f X (x
( x) =
0
otherwise

Determine the cumulative distribution function of flange

thickness.
Determine the proportion of flanges that exceeds 1.02 mm.
What thickness is exceeded by 90% of the flanges?
Determine the mean and variance of flange thickness.
Ignacio Cascos

5. Random vectors
25
Ignacio Cascos
Transformations of random variables
Given a random variable X and a function

g: IR IR, we want to describe the distribution
of the random varaible Y=g(X).
If X is a discrete random variable,
FY(y) = P(Y y) = P(g(X) y) = P(X Ay) ,

where Ay = {x: g(x) y}.
It often happens that the set Ay can be fully
described in a simple way.
Ignacio Cascos
27
26
FY(y) = P(Y y) = P(g(X) y)

= g(xi) y pX(xi) .
Further, the probability mass function of Y is
pY(y) = P(Y = y) = P(g(X) = y)

= g(xi) = y pX(xi) .
Ignacio Cascos
28
Outline
If g is continuous and increasing

FY(y) = P(g(X) y) = P(X g1(y)) = FX(g1(y))

In general, given X continuous r.v. and Y=g(X)
with g injective and derivable, the density mass
function of Y satisfies



5. Random vectors
fY (y) = fX (x) |dx/dy|

Ignacio Cascos
29
Ignacio Cascos
30
Joint distribution of a random vector
Discrete random vectors
Given X and Y two random variables in the same

sample space E, the mapping
(X,Y) : E IR2
is a bivariate random vector.
The joint cumulative distribution function of
(X,Y) at (x,y) is given by
Given two discrete random variables X and Y,

the random vector (X,Y) is discrete and its joint
probability mass function is given by
p(x,y) = P(X = x , Y = y).
It satisfies:
FX,Y(x,y) = P(X x , Y y) = P((X x)(Y y))
1.
2.
p(x,y) 0
xyp(x,y) = 1
Joint cumulative distribution function

F(x0,y0) = P(X x0 , Y y0) = xx yy p(x,y)
0
Ignacio Cascos
31
Ignacio Cascos
32
Continuous random vectors
Marginal distributions
For X and Y continuous rvs, the random vector (X,Y)

is continuous and its joint density function f(x,y) satisfies
1.
f(x,y) 0 ;
2.
f(x,y)dxdy = 1 .
The distribution of each of the components of

a random vector alone is referred to as
marginal distribution.
Joint cumulative distribution function

F(x0,y0) = P(X x0 , Y y0) = f(x,y)dydx
x0
y0
That satisifes
P(a X b , c Y d) = abcd f(x,y)dydx
f(x,y) = 2F(x,y)/xy
Ignacio Cascos
33
Marginal distributions
Ignacio Cascos
Independent random variables
Discrete variables. Given X and Y whose joint

probability mass function is denoted as p(x,y),
Two random variables X and Y are independent

if for any A,BIR,
pX(x) = P(X=x) = yP(X=x,Y=y) = yp(x,y)

pY(y) = P(Y=y) = xP(X=x,Y=y) = xp(x,y)
P((XA)(YB)) = P(XA)P(YB)
Equivalently, for any x,yIR
Continuous variables. Given X and Y whose joint

density mass function is denoted as f(x,y),
+
fX(x) = f(x,y)dy
+
fY(y) = f(x,y)dx
P((X x)(Y y)) = P(X x)P(Y y)

Property. If X and Y are independent,
Var[X+Y] = Var[X]+Var[Y]
Ignacio Cascos
34
35
Ignacio Cascos
36
Characteristic features of a rand. vector
Given a function g: IR2 IR, the expectation of

random vector g(X,Y) can be computed as
Expectation. The mean vector of a random vector X

is a column vector each of whose entries is the
expectation of an entry of X
E[ X 1 ]
E[ X 2 ]
E[ X ] =
M
E[ X ]
n
Ignacio Cascos
(X,Y) discrete, E[g(X,Y)] = g(xi,yj)p(xi,yj)

(X,Y) cont., E[g(X,Y)] = g(x,y)f(x,y)dxdy
37

Covariance. The covariance is a measure of
the linear relation between two variables,
Cov(X,Y) = E[(XE[X])(YE[Y])]
Properties:
1.
2.
3.
4.
Ignacio Cascos
Ignacio Cascos
38

Correlation. The correlation is also a measure
of the linear relation between two variables
( X ,Y ) =
Cov(X,Y) = E[XY]E[X]E[Y] ;
if X and Y are independent, Cov(X,Y) = 0 ;
Cov(X,Y) = 0 does not guarantee that X and Y are
independet ;
Cov(aX+b,cY+d)=acCov(X,Y) .
39
Properties:
1.
2.
3.
4.
Ignacio Cascos
Cov( X , Y )
Var[ X ]Var[Y ]
if X and Y are independent, (X,Y) = 0 ;

if (X,Y) = 0 they are said to be uncorrelated ;
| (X,Y) | 1 ;
| (X,aX+b) | = 1 .
40
A dartboard is a wooden square of 40cmx40cm. Inside the board

there are several concentric circles, the greater among them
having a diameter of 40cm. The darts of an unskilled player hit
the board at a point that can be described by a random vector
(X,Y) with joint density function
Covariance matrix.
M X = E[( X )( X ) t ]
Cov( X 1 , X 2 )
Var[ X 1 ]
Var[ X 2 ]
Cov( X 2 , X 1 )
MX =
M
M
Cov( X , X ) Cov( X , X )
n
1
n
2
Ignacio Cascos
L Cov( X 1 , X n )
L Cov( X 2 , X n )
O
M
L
Var[ X n ]
1 / 1600 if 0 x, y 40
f X ,Y ( x, y ) =
otherwise
0
where (X,Y) represent the coordinates (in centimeters) of the
hitting point. The lower right corner of the board is represented
by the origin, that is, the point (0,0).
41
Example (Dartboard)
Whenever we hit one of the circles, we get some
point. What is the probability of getting some point
from a throw?
What is the mean vector of (X,Y)?, and the
probability that (X,Y) assumes that value?
The inner circle has a diameter of 4cm. What is the
probability of hitting it with a dart?
Ignacio Cascos
Example (Dartboard)
43
Ignacio Cascos
42

2rvariables Slides

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2rvariables Slides

Uploaded by

Copyright:

Available Formats

Outline

1. Definition of random variable

3. Characteristic features of a random variable

4. Transformations of random variables

Depto. Estadstica, Universidad Carlos III

Definition of random variable

Depto. Estadstica, Universidad Carlos III

Definition of random variable

A random variable assesses a real number to

Definition. A random variable X is a mapping

It is random because we do not know its value

Depto. Estadstica, Universidad Carlos III

Depto. Estadstica, Universidad Carlos III

Definition of random variable

The events are now of the type XA,

Discrete random variables (probability mass function and cdf)

Their probabilities are P(XA) = P({eE: X(e)A}).

3. Characteristic features of a random variable

4. Transformations of random variables

if A1, A2,IR satisfy AiAj= for i j,

Joint distribution and independence

then P(Xi=1, Ai)=i=1, P(XAi) .

Depto. Estadstica, Universidad Carlos III

Discrete and continuous random variables

Examples: number of defective items, number of rolls of a

A random variable is continuous if its range contains

Depto. Estadstica, Universidad Carlos III

Discrete random variables

The range of a random variable is the set of values that

Depto. Estadstica, Universidad Carlos III

Discrete random variables

Discrete random variables

Let X be the number of heads

The cumulative distribution function (cdf) of X

numero motores averiados

Depto. Estadstica, Universidad Carlos III

Depto. Estadstica, Universidad Carlos III

Continuous random variables

Since the range of a continuous random variable is not

Histogram for the lifetime of 10000 batteries.

numero motores averiados

Discrete random variables

Depto. Estadstica, Universidad Carlos III

Depto. Estadstica, Universidad Carlos III

Function f describes the

The density mass function f describes the probability

Continuous random variables

Given X continuous r.v., it satisfies

Continuous random variables

Depto. Estadstica, Universidad Carlos III

Depto. Estadstica, Universidad Carlos III

Continuous random variables

Continuous random variables

In order to compute the cumulative distribution

Since the cumulative distribution

Depto. Estadstica, Universidad Carlos III

We are working with

Depto. Estadstica, Universidad Carlos III

Mathematical expectation or mean

1. Definition of random variable

The expecation or mean () of a random variable

Discrete random variables (probability mass function and cdf)

X discrete, = E[X] = xip(xi)

3. Characteristic features of a random variable