0% found this document useful (0 votes)
34 views

Topic 2 Probability Theory-Part A

The document discusses key concepts in probability and statistics including random variables, probability distributions, expectation, variance, and relationships between multiple random variables. It defines discrete and continuous random variables and their probability distributions. It also covers topics like expected value, variance, marginal and conditional probability distributions, and independence of random variables.

Uploaded by

daddy's cock
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
34 views

Topic 2 Probability Theory-Part A

The document discusses key concepts in probability and statistics including random variables, probability distributions, expectation, variance, and relationships between multiple random variables. It defines discrete and continuous random variables and their probability distributions. It also covers topics like expected value, variance, marginal and conditional probability distributions, and independence of random variables.

Uploaded by

daddy's cock
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 40

A quick review of calculus Econ 3334

 Use calculus to find Min or Max of functions


 Multi-value functions
z = −5x 2 + 10xy − 20x − 7y 2 + 240y − 5300
 Find the x & y that gives the Max z
∂z ∂z
∂x
= 0 and ∂y
= 0 are the first order conditions (FOCs)
 Solve simultaneously for x & y
∂z
∂x
= −10x + 10y − 20 = 0
∂z
∂y
= 10x − 14y + 240 = 0
 Solve simultaneously: y ∗ = 55 x ∗ = 53
 Plug in x* and y* into z to get the value z ∗ = 770
 Is it a maximum or minimum? Need 2nd order conditions (don’t
1
worry here). Here it is the maximum.
Summations Econ 3334

 We’ll use summation notation a lot


If x = 8, −3, 4, 2, −1, n = 5
then ∑ i=1
n
x i = 8 + −3 + 4 + 2 + −1 = 10
If y = 0, 0, 10, −1, −1, then ∑ y = 8
If a is constant, then ∑ i=1
n
a =a + a + a +. . . . . +a= na
n times

e.g. ∑ i=1
99
10 =10 + 10 +. . . . . +10= 99 ⋅ 10 = 990
99 times
 Note: ∑ x i ⋅ ∑ y i
= 10 ⋅ 8
≠ ∑ x i ⋅ yi
= 8 ⋅ 0 + −3 ⋅ 0 + 4 ⋅ 10 + 2 ⋅ −1 + −1 ⋅ −1 = 39
Summations Econ 3334

 A few important properties

∑ ax i = a ∑ x i
∑a = n ⋅ a
∑ xiyi = ∑ yixi
∑ax i + by i  = a ∑ x i + b ∑ y i
∑ ∑ xi = n ∑ xi
Econ 3334

Review of Probability Theory

4
1) Random variable and distribution Econ 3334
 Outcomes- Potential results
 Sample Space- All possible outcomes
 Probability- Proportion of the time that an outcome occurs
 Event- Set of one or more outcomes (subset of sample space)
 e.g. Grade Probability
A 4.0 .15
B 3.0 .40
C 2.0 .30
D 1.0 .10
F 0.0 .05
 A,B,C,D,F are each outcomes, together they are sample
space
 ∑ prob = 1 and each prob∈ 0, 1 
1) Random variable and distribution Econ 3334
 Let V=the event that a random student gets an A or B
In this example “Grade” is a discrete random variable
PV = PA + PB  =. 15 +. 40 =. 55

 Two types of random variables (RV):


• Discrete random variable: takes only a discrete set
of values, like 0,1,2…,
• Continuous random variable: takes on a continuum
of possible values.
1) Random variable and distribution Econ 3334

 Probability Distribution
For a discrete random variable, the probability distribution
lists all the probabilities (e.g. our grade example)

Outcomes x 1 x 2 . . . x k
Probability p 1 p 2 . . . p k
1) Random variable and distribution Econ 3334
 For continuous random variable, we have an
equation called probability density function that
does a similar thing. It provides the probability
that the RV takes on a value within an interval.

 E.g. Normal Dist: x−μ 2


− 2
φx;μ, σ  = 1
e 2σ
2π σ
2
• When μ = 0, σ = 1 , φx = 1 − x2
e

1) Random variable and distribution Econ 3334

 Normal probability density


1) Random variable and distribution Econ 3334

 CDF (cumulative density function)


 If X is a random variable, then its CDF is defined
for any real number x by
Fx = PrX ≤ x
 So, F−∞ = 0; F∞ = 1 , and F takes values between 0
and 1
 If a<b, then F(a) ≤ F(b), (that is, F is non-decreasing
in its argument)
 If a<b, then Pr(a<X<b)=F(b)-F(a)
1) Random variable and distribution Econ 3334

11
Econ 3334

12
2) Expectation, Mean, and Variance Econ 3334

 Expected value of RV X is the average value of the RV


over many repeated trials.
EX = μ X = mean of X
 For discrete random variables:
EX = μ X = ∑ i x i ⋅ px i 
 Grade Probability
A 4.0 .15
EX = ∑ x i ∗ Px i 
B 3.0 .40
C 2.0 .30 = 4. 0 ∗. 15 + 3. 0 ∗. 40 +. . . . . +0 ∗. 05

D 1.0 .10 = 2. 5

F 0.0 .05
2) Expectation, Mean, and Variance Econ 3334

 For a continuous variable

EX  = μ X = ∫ x ⋅ fx dx,


where fx  is the probability density function (PDF).
2) Expectation, Mean, and Variance Econ 3334

 Variance & Standard deviation are measures of “spread”


varX = σ 2X = E X − μ X  2 = EX 2  − μ 2X
 The standard deviation is simply
σX = σ 2X
 For a discrete RV,
varX  = σ 2X = ∑x i − μX  2 ⋅ px i 
i
 For a continuous RV,

varX = ∫x − μ X  2 ⋅ fxdx


2) Expectation, Mean, and Variance Econ 3334

 Example:
Grade Probability
A 4.0 .15
σ 2X = 4 − 2.5 2 ⋅ 0.15 + 3 − 2.5 2 ⋅ 0.40 +...
B 3.0 .40
C 2.0 .30 = 1.05
D 1.0 .10
F 0.0 .05 σ X = 1. 05 = 1. 025
2) Expectation, Mean, and Variance Econ 3334

 Suppose
Y = a + bX
EY = Ea + bX = a + bEX
VarY = Vara + bX = 0 + b 2 VarX

 E.g. Consumption & Income

C = 1000 + 0. 9I EI = 4000 VarI = 800

 Then
EC = 1000 + 0. 9EI = 1000 + 0. 9 ⋅ 4000 = 4600

varC = 0. 9 2 varI = 0. 81 ⋅ 800 = 648


2) Expectation, Mean, and Variance Econ 3334

 Measures of Shape of Dist


1)Mean (Central tendency)
2)Variance & Standard deviation (Dispersion)
3) skewness = measure of asymmetry of a distribution
E X−μ X  3
skewness =
σ 3X

 Skewness = 0: distribution is symmetric


 Skewness > (<) 0: distribution has long right (left) tail
2) Expectation, Mean, and Variance Econ 3334

 4) kurtosis = measure of mass in tails


= measure of probability of large values

E X−μ X  4
kurtosis =
σ 4X

 kurtosis = 3: normal distribution


 kurtosis > 3: heavy tails (“leptokurtotic”)
2) Expectation, Mean, and Variance Econ 3334


3) Two Random Variables Econ 3334

 Consider 2 RVs X & Y


e.g. X is Income, Y is Gender
e.g. X is Parent height , Y is Child’s height
e.g. X is Price, Y is Qty demand

 Joint Prob Dist

 Marginal Dist

 Conditional Dist
3) Two Random Variables Econ 3334
 Joint distribution : Random variables X and Y have a joint
distribution, i.e., the probability that the random variables
simultaneously take on certain values, say x and y.
 The time to commute to school, and the weather
Time\weather 0 (raining) 1 (not raining)
40 min 0.08 0.15
50 min 0.20 0.25
60 min 0.28 0.04

 Joint “and”
PX = x, Y = y
e. g PX = 60, Y = 0 =. 28
3) Two Random Variables Econ 3334
 Marginal probability distribution of a random variable Y is
just another name for its probability distribution.
 This term is used to distinguish the distribution of Y alone
and another random variable, say X.
 PY = y = ∑ i PX = x i, Y = y
PY = 0 = ∑ i PX = x i, Y = 0 = 0. 56

 Time\weather 0 (raining) 1 (not raining) Marginal of time


40 min 0.08 0.15 0.23
50 min 0.20 0.25 0.45
60 min 0.28 0.04 0.32
Marginal of weather 0.56 0.44 1
3) Two Random Variables Econ 3334

 Conditional Distribution:
PX=x,Y=y
PX = x ∣ Y = y = PY=y
PX=x,Y=y
PY = y ∣ X = x = PX=x

• What is the probability of X=60 given that Y=1?


PX = 60 ∣ Y = 1 = 0.04
0.44
= 0. 09

• What is the probability of Y=1 given that X=60?

PY = 1 ∣ X = 60 = 0.04


0.32
= 0. 125
3) Two Random Variables Econ 3334

Time\weather 0 (raining) 1 (not raining) Marginal of time


40 min 0.08 0.15 0.23
50 min 0.20 0.25 0.45
60 min 0.28 0.04 0.32
Marginal of weather 0.56 0.44 1

Time Condition distribution Time Condition distribution


given whether =0 given whether =1
40 min 0.08/0.56=0.143 40 min 0.15/0.44=0.340
50 min 0.20/0.56=0.357 50 min 0.25/0.44=0.568
60 min 0.28/0.56=0.5 60 min 0.04/0.44=0.091
sum 1 sum 1
3) Two Random Variables Econ 3334

 X & Y are independent if


PX = x ∣ Y = y = PX = x
for all the possible x and y.
 i.e. “given” Y does not change the probability
 Since Px=x,Y=y
PX = x ∣ Y = y = PY=y

then X & Y are independent of

PX = x, Y = y = PX = x ⋅ PY = y


for all the possible x and y.
3) Two Random Variables Econ 3334
 Let’s check our time/weather example
Time\weather 0 (raining) 1 (not raining) Marginal of time
40 min 0.08 0.15 0.23
50 min 0.20 0.25 0.45
60 min 0.28 0.04 0.32
Marginal of weather 0.56 0.44 1

 PX = 60 ∣ Y = 1 = 0. 09
PX = 60 = 0. 32
0. 09 ≠ 0. 32
so clearly X & Y are NOT independent
3) Two Random Variables Econ 3334
 Conditional Expectation is the mean of Y given X=x:
EY ∣ X = x  = ∑ y i ⋅ PY = y i ∣ X = x 

 Conditional Expectation is the mean of X given Y=y:

EX ∣ Y = y = ∑ x i ⋅ PX = x i ∣ Y = y

 If X and Y are independent, then


EY ∣ X = EY and EX ∣ Y = EX
3) Two Random Variables Econ 3334

 Time/weather example
Time Condition distribution Time Condition distribution
given whether =0 given whether =1
40 min 0.08/0.56=0.143 40 min 0.15/0.44=0.340
50 min 0.20/0.56=0.357 50 min 0.25/0.44=0.568
60 min 0.28/0.56=0.5 60 min 0.04/0.44=0.091
sum 1 sum 1

EX ∣ Y = 0 = 40 ⋅ 0.08
0.56
+ 50 ⋅ 0.20
0.56
+ 60 ⋅ 0.28
0.56
= 53. 57

EX ∣ Y = 1 = 40 ⋅ 0.15
0.44
+ 50 ⋅ 0.25
0.44
+ 60 ⋅ 0.04
0.44
= 47. 5
3) Two Random Variables Econ 3334

 Suppose that X is a random variable:


x 1 with probability p1
x 2 with probability p2
X=
...
x k with probability pk


Suppose that Y is a random variable:
y 1 with probability q1
y 2 with probability q2
Y=
...
y m with probability qm
3) Two Random Variables Econ 3334

 Define a random variable EY ∣ X as:


EY ∣ X = x 1  with probability p 1
EY ∣ X = x 2  with probability p 2
EY ∣ X =
...
EY ∣ X = x k  with probability p k
This is just how we define the random variable EY ∣ X

 Define a random variable EX ∣ Y as


EX ∣ Y = y 1  with probability q 1
EX ∣ Y = y 2  with probability q 2
EX ∣ Y =
...
EX ∣ Y = y m  with probability q m
This is just how we define the random variable EX ∣ Y
3) Two Random Variables Econ 3334

 Law of Iterated Expectations (LIE):


EY = EEY ∣ X

E EY ∣ X
take the expectation w ith respect to Y given X

take the expectation w ith respect to X

 EX = EEX ∣ Y


 In the discrete case
EX = ∑EX ∣ Y = y i  ⋅ pY = y i 
EY = ∑EY ∣ X = x i  ⋅ pX = x i 
3) Two Random Variables Econ 3334
 Let’s check our time/weather example
Time\weather 0 (raining) 1 (not raining) Marginal of time
40 min 0.08 0.15 0.23
50 min 0.20 0.25 0.45
60 min 0.28 0.04 0.32
Marginal of weather 0.56 0.44 1

 EX = ∑ x ⋅ PX = x  = 40 ⋅ 0. 23 + 50 ⋅ 0. 45 + 60 ⋅ 0. 32 = 50. 9


i i

 We show: EX ∣ Y = 0 = 40 ⋅ 0.08 0.56


+ 50 ⋅ 0.20
0.56
+ 60 ⋅ 0.28
0.56
= 53. 57

EX ∣ Y = 1 = 40 ⋅ 0.15
0.44
+ 50 ⋅ 0.25
0.44
+ 60 ⋅ 0.04
0.44
= 47. 5

 By LIE, EX = ∑ EX ∣ Y = yi  ⋅ PY = yi 


= EX ∣ Y = 0 ⋅ pY = 0 + EX ∣ Y = 1 ⋅ pY = 1
= 53. 57 ⋅ 0. 56 + 47. 5 ⋅ 0. 44
= 50. 9
3) Two Random Variables Econ 3334
 Properties of conditional expectation
 Ec|X = c, where c is a constant
 EaY + b|X = aEY|X + b
 EfX|X = fX
 Ef 1 X ⋅ f 2 Y|X = f 1 X ⋅ Ef 2 Y|X
 EY|X = EY if X and Y are independent
 EY = EEY|X (law of iterated expectation)
 Example EU|X = 0 implies that EU = 0 and EUX = 0
Proof: EU = EEU|X = E0 = 0 (law of iterated expectation)
EUX = EEUX|X (law of iterated expectation)
= EX ⋅ EU|X = EX ⋅ 0 = 0 ■
3) Two Random Variables Econ 3334
 Conditional Variance:
VarY ∣ X = x  = E Y − EY ∣ X = x  2 X = x
VarX ∣ Y = y = E X − EX ∣ Y = y 2 Y = y
 Discrete case:
varY ∣ X = x = ∑ y i − EY ∣ X = x 2 ⋅ PY = y i ∣ X = x 
varX ∣ Y = y = ∑ x i − EX ∣ Y = y 2 ⋅ PX = x i ∣ Y = y

 If X and Y are independent


varY ∣ X = x  = varY
varX ∣ Y = y = varX
3) Two Random Variables Econ 3334

 Conditional Variance:
Time Condition distribution
given whether =0
40 min 0.08/0.56=0.143
50 min 0.20/0.56=0.357
60 min 0.28/0.56=0.5
sum 1

EX ∣ Y = 0 = 40 ⋅ 0.08
0.56
+ 50 ⋅ 0.20
0.56
+ 60 ⋅ 0.28
0.56
= 53. 57

varX ∣ Y = 0
= 40 − 53. 57 2 ⋅ 0. 08 + 50 − 53. 57 2 0. 20 + 60 − 53. 57 2 0. 28
0. 56 0. 56 0. 56
= 51. 53
3) Two Random Variables Econ 3334

 Covariance: (linear) statistical relationship between


variables

covX, Y = σ XY = EX − μ X Y − μ Y  = EXY − μ X μ Y

 Correlation:
covX,Y σ XY
corrX, Y = ρ XY = = σ X ⋅σ Y
varX  varY


− 1 ≤ corr ≤ 1

 If X and Y are independent, then corr(X,Y)=0


3) Two Random Variables Econ 3334

 corr(X,Y) = 1 mean perfect positive linear association

 corr(X,Y) = –1 means perfect negative linear association

 corr(X,Y) = 0 means no linear association


3) Two Random Variables Econ 3334
The correlation coefficient measures linear association
3) Two Random Variables Econ 3334

 Mean and variance of sums of RVs


EaX + bY = aEX + bEY

varaX + bY = a 2 varX + b 2 varY + 2ab ⋅ covX, Y


 If X and Y are independent
varaX + bY = a 2 varX + b 2 varY

You might also like