jane bae - stanford universitylanhuong/refresher/notes/prob... · 2015-05-31 · overview 1...
TRANSCRIPT
![Page 1: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/1.jpg)
Probability and Statistics
Jane Bae
Stanford University
September 16, 2014
Jane Bae (Stanford) Probability and Statistics September 16, 2014 1 / 35
![Page 2: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/2.jpg)
Overview
1 Probability ConceptsProbability SpaceCombinatoricsRandom Variables and DistributionsIndependence and Conditional Probability
Jane Bae (Stanford) Probability and Statistics September 16, 2014 2 / 35
![Page 3: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/3.jpg)
Outline
1 Probability ConceptsProbability SpaceCombinatoricsRandom Variables and DistributionsIndependence and Conditional Probability
Jane Bae (Stanford) Probability and Statistics September 16, 2014 3 / 35
![Page 4: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/4.jpg)
Probability Space
A probability space consists of three parts:
A sample space, Ω,
A set of events F ,
A probability P.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 4 / 35
![Page 5: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/5.jpg)
Sample Space
A sample space is the set of all possible outcomes.
Single coin flip:Ω = H,T.
Double coin flip:
Ω = (H,H), (H,T ), (T ,H), (T ,T ).
Single dice roll:Ω = 1, 2, 3, 4, 5, 6.
Double dice roll:Ω =?
Jane Bae (Stanford) Probability and Statistics September 16, 2014 5 / 35
![Page 6: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/6.jpg)
Set of Events
A event is a subset of the sample space.
Single coin flip: (Event that a coin lands head)
A = H.
Double coin flip: (Event that first coin lands head)
A = (H,H), (H,T ).
Single dice roll: (Event that the dice roll is less than 4)
A = 1, 2, 3.
Double dice roll: (Event that the dice roll add up to less than 4)
A = (1, 1), (1, 2), (2, 1)
Jane Bae (Stanford) Probability and Statistics September 16, 2014 6 / 35
![Page 7: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/7.jpg)
The collection of all such events we would like to consider are calledthe set of events, F .
F is a σ-algebra.
Ω ∈ F .F is closed under complementation.F is closed under countable unions.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 7 / 35
![Page 8: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/8.jpg)
Probability
A set function on the set of events such that
0 ≤ P(A) ≤ 1 for all events A.
P(Ω) = 1.
For each sequence of mutually disjoint events A1,A2,A3, ...,
P
( ∞⋃i=1
Ai
)=∞∑i=1
P(Ai ).
(countable additivity).
Jane Bae (Stanford) Probability and Statistics September 16, 2014 8 / 35
![Page 9: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/9.jpg)
Probability
Fair coin
P(∅) = 0
P(H) = 1/2
P(T) = 1/2
P(H,T) = 1
Fair die
P(∅) = 0
P(1) = 1/6
P(2) = 1/6
P(1, 2, 3) = 1/2
Jane Bae (Stanford) Probability and Statistics September 16, 2014 9 / 35
![Page 10: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/10.jpg)
Some Properties (Good Exercise Problems)
P(∅) = 0.
P(Ac) = 1− P(A).
If A ⊆ B, then P(A) ≤ P(B).
P(A ∪ B) = P(A) + P(B)− P(A ∩ B).
For each sequence of events A1,A2,A3, ... (not necessarily disjoint),
P
( ∞⋃i=1
Ai
)≤∞∑i=1
P(Ai ).
Jane Bae (Stanford) Probability and Statistics September 16, 2014 10 / 35
![Page 11: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/11.jpg)
Continuous Probability Space
Similar idea as discrete probability space.
Example
Arrival time of trains passing with approxmiately equal intervals.Chord length of a circle of radius R with chord randomly selected.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 11 / 35
![Page 12: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/12.jpg)
Probability Space
A probability space consists of three parts:
A sample space, Ω,
A set of events F ,
A probability P.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 12 / 35
![Page 13: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/13.jpg)
Sample Space
A sample space is the set of all possible outcomes.
Random number between 0 and 1:
Ω = [0, 1).
Height of 10-year-olds (in cm):
Ω = [0,∞).
Jane Bae (Stanford) Probability and Statistics September 16, 2014 13 / 35
![Page 14: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/14.jpg)
Set of Events
A event is a subset of the sample space.
Random number between 0 and 1: (Event that it is less than 0.5)
A = [0, 0.5)
Height of 10-year-olds (in cm): (Event that a 10-year-old is tallerthan 150cm)
A = [150,∞)
Jane Bae (Stanford) Probability and Statistics September 16, 2014 14 / 35
![Page 15: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/15.jpg)
Outline
1 Probability ConceptsProbability SpaceCombinatoricsRandom Variables and DistributionsIndependence and Conditional Probability
Jane Bae (Stanford) Probability and Statistics September 16, 2014 15 / 35
![Page 16: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/16.jpg)
Permutations
Permutation relates to the act of rearranging members of a set into aparticular sequence or order.
Ex) Six permutations of the set 1,2,3Ex) Ways to sit around a table
k-permutation of n are the different ordered arrangements of a k-elementsubset of an n-set, usually denoted by
nPk =n!
(n − k)!.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 16 / 35
![Page 17: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/17.jpg)
Combinations
Combination is a way of selecting members from a grouping, such thatthe order of selection does not matter.
Ex) Drawing marbles out of a box
k-combination of n are the different groupings of a k-element subset ofan n-set, usually denoted by
nCk =n!
k!(n − k)!.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 17 / 35
![Page 18: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/18.jpg)
Outline
1 Probability ConceptsProbability SpaceCombinatoricsRandom Variables and DistributionsIndependence and Conditional Probability
Jane Bae (Stanford) Probability and Statistics September 16, 2014 18 / 35
![Page 19: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/19.jpg)
Random Variables
A random variable is a function from a sample space to a real number.
X : Ω→ R
Example:
Single coin flip
X (H) = 1,
X (T ) = −1.
Double dice roll
X (i , j) = i
Y (i , j) = j
Z (i , j) = i + j .
Jane Bae (Stanford) Probability and Statistics September 16, 2014 19 / 35
![Page 20: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/20.jpg)
Distributions of Discrete Random Variables
A discrete random variable X assumes values in discrete subset S of R.The distribution of a discrete random variable is completely described by aprobability mass function pX : R→ [0, 1] such that
P(X = x) = pX (x).
Example:
Bernoulli: X ∼ Ber(p) if X ∈ 0, 1 and
pX (1) = 1− pX (0) = p.
Binomial: X ∼ Bin(n, p) if X ∈ 0, 1, ..., n and
pX (k) =
(n
k
)pk(1− p)n−k .
Poisson: X ∼ Pois(λ) if X ∈ 0, 1, ... and
pX (k) =λke−λ
k!.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 20 / 35
![Page 21: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/21.jpg)
Distributions of Continuous Random Variables
A continuous random variable X assumes values in R.The distribution of continuous random variables is completely described bya probability density function fX : R→ R+ such that
P(a ≤ X ≤ b) =
∫ b
afX (x)dx .
Example:
Uniform: X ∼ U(a, b), a ≤ b if
fX (x) =
1
b−a , a ≤ x ≤ b
0, otherwise.
Chi-Square: Q ∼ χ2(k), k ∈ Z if
fX (x) =
xk/2−1e−x/2
2k/2Γ(k/2), x ≥ 0
0, otherwise.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 21 / 35
![Page 22: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/22.jpg)
Gaussian/Normal: X ∼ N (µ, σ2), µ ∈ R, σ2 > 0 if
fX (x) =1√
2πσ2e−
(x−µ)2
2σ2
Jane Bae (Stanford) Probability and Statistics September 16, 2014 22 / 35
![Page 23: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/23.jpg)
Probability Distribution
Includes the probability mass function and probability density functionmentioned above.Also includes:
Cumulative distribution function
FX (x) = P(X ≤ x), x ∈ R.
Characteristic function
Any rule that defines a distribution
NOTATION: The right-hand sides of the previous displays are shorthandnotation for the following:
P(X ≤ x) = P(ω ∈ Ω : X (ω) ≤ x)
Jane Bae (Stanford) Probability and Statistics September 16, 2014 23 / 35
![Page 24: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/24.jpg)
Note 1: Distribution can be identical even if the supporting probabilityspace is different. Example:
X (H) = 1
X (T ) = −1
Y (i) =
1, i is odd
−1, i is even
Note 2: Distribution can be different even if the supporting probabilityspace is identical.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 24 / 35
![Page 25: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/25.jpg)
Joint Distribution
Two random variables X and Y induce a probabillity PX ,Y on R2:
PX ,Y ((−∞, x ]× (−∞, y ]) = P(X ≤ x ,Y ≤ y).
A collection of random variables X1,X2, ...,Xn induce a probabillityPX1,...,Xn on Rn:
PX1,...,Xn((∞, x1]× · · · × (∞, xn]) = P(X1 ≤ x1, · · · ,Xn ≤ xn).
Jane Bae (Stanford) Probability and Statistics September 16, 2014 25 / 35
![Page 26: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/26.jpg)
Joint distribution of two discrete random variables X and Y assumingvalues in SX and SY can be completely described by joint probability massfunction pX ,Y : R× R→ [0, 1] such that
P(X = x ,Y = y) = pX ,Y (x , y).
Joint distribution of two continuous random variables X and Y can becompletely described by joint probability density functionfX ,Y : R× R→ R+ such that
P(X ≤ x ,Y ≤ y) =
∫ x
−∞
∫ y
−∞fX ,Y (x , y)dydx .
Jane Bae (Stanford) Probability and Statistics September 16, 2014 26 / 35
![Page 27: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/27.jpg)
Outline
1 Probability ConceptsProbability SpaceCombinatoricsRandom Variables and DistributionsIndependence and Conditional Probability
Jane Bae (Stanford) Probability and Statistics September 16, 2014 27 / 35
![Page 28: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/28.jpg)
Independence
Two events A and B are independent if and only if their joint probabilityequals the product of their probabilities:
P(A ∩ B) = P(A)P(B).
A finite set of events Ai is pairwise independent iff every pair of eventsis independent. That is, if and only if for all distinct pairs of indices m, n
P(Am ∩ An) = P(Am)P(An).
A finite set of events is mutually independent if and only if every event isindependent of any intersection of the other events.That is, iff for everysubset An
P
(n⋂
i=1
Ai
)=
n∏i=1
P(Ai ).
Jane Bae (Stanford) Probability and Statistics September 16, 2014 28 / 35
![Page 29: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/29.jpg)
A set of random variables is pairwise independent iff every pair of randomvariables is independent.That is either joint cumulative distribution function
FX ,Y (x , y) = FX (x)FY (y),
or equivalently, joint density
fX ,Y (x , y) = fX (x)fY (y).
A set of random variables is mutually independent iff for any finite subsetX1, . . . ,Xn and any finite sequence of numbers a1, . . . , an, the eventsX1 ≤ a1, . . . , Xn ≤ an are mutually independent events.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 29 / 35
![Page 30: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/30.jpg)
Conditional Probability
The conditional probability of A given B is defined as
P(A|B) =P(A ∩ B)
P(B).
Jane Bae (Stanford) Probability and Statistics September 16, 2014 30 / 35
![Page 31: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/31.jpg)
Conditional Probability Mass and Density
If X and Y are both discrete random variables with joint probability massfunction pX ,Y (x , y),
P(X = x |Y = y) = pX |Y (x |y) :=pX ,Y (x , y)
pY (y).
If X and Y are both continuous random variables with joint densityfunction fX ,Y (x , y),
P(a ≤ X ≤ b|Y = y) =
∫ a
bfX |Y (x |y)dx ,
where
fX |Y (x |y) =fX ,Y (x , y)
fY (y).
Jane Bae (Stanford) Probability and Statistics September 16, 2014 31 / 35
![Page 32: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/32.jpg)
Bayes’ theorem
The relationship between P(A|B) and P(B|A) is given by
P(B|A) =P(A|B)P(B)
P(A).
That is, P(A|B) = P(B|A) only if P(B)/P(A) = 1, or equivalently,P(A) = P(B).
Jane Bae (Stanford) Probability and Statistics September 16, 2014 32 / 35
![Page 33: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/33.jpg)
Exercise
Consider a family with two children. Given that one of the children isa boy, what is the probability that both children are boys?
Mr. Smith is the father of two. We meet him walking along the streetwith a young boy whom he proudly introduces as his son. What is theprobability that Mr. Smiths other child is also a boy?
Jane Bae (Stanford) Probability and Statistics September 16, 2014 33 / 35
![Page 34: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/34.jpg)
From 5 August 2011 New York Times article by John Allen Paulos:”Assume that you are presented with three coins, two of them fair and theother a counterfeit that always lands heads. If you randomly pick one ofthe three coins, the probability that it’s the counterfeit is 1 in 3. This isthe prior probability of the hypothesis that the coin is counterfeit. Nowafter picking the coin, you flip it three times and observe that it landsheads each time. Seeing this new evidence that your chosen coin haslanded heads three times in a row, you want to know the revised posteriorprobability that it is the counterfeit. The answer to this question, foundusing Bayes’s theorem (calculation mercifully omitted), is 4 in 5. You thusrevise your probability estimate of the coin’s being counterfeit upward from1 in 3 to 4 in 5.”
Jane Bae (Stanford) Probability and Statistics September 16, 2014 34 / 35
![Page 35: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/35.jpg)
Suppose a drug test is 99% sensitive. That is, the test will produce 99%true positive results for drug users. Suppose that 0.5% of people are usersof the drug. If a randomly selected individual tests positive, what is theprobability he or she is a user?
Jane Bae (Stanford) Probability and Statistics September 16, 2014 35 / 35
![Page 36: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/36.jpg)
Probability and Statistics
Jane Bae
Stanford University
September 15, 2014
Jane Bae (Stanford) Probability and Statistics September 15, 2014 1 / 39
![Page 37: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/37.jpg)
Overview
1 Probability ConcetpsExpectation and MomentsConditional ExpectationMoment Generating Function, Characteristic Function
2 Limit TheoremsModes of ConvergenceLaw of Large NumbersCentral Limit Theorem
3 StatisticsEstimatorsEstimation Strategies
Jane Bae (Stanford) Probability and Statistics September 15, 2014 2 / 39
![Page 38: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/38.jpg)
Outline
1 Probability ConcetpsExpectation and MomentsConditional ExpectationMoment Generating Function, Characteristic Function
2 Limit Theorems
3 Statistics
Jane Bae (Stanford) Probability and Statistics September 15, 2014 3 / 39
![Page 39: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/39.jpg)
Expectation
For discrete random variable X , the expectation of X is
E [X ] =∑x∈S
xpX (x).
For continuous random variable Y , the expectation of Y is
E [Y ] =
∫ ∞−∞
yfY (y)dy .
Jane Bae (Stanford) Probability and Statistics September 15, 2014 4 / 39
![Page 40: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/40.jpg)
We can also compute expecation of g(X ) and g(Y ) as
E [g(X )] =∑x∈S
g(x)pX (x),
and
E [g(Y )] =
∫ ∞−∞
g(y)fY (y)dy .
Jane Bae (Stanford) Probability and Statistics September 15, 2014 5 / 39
![Page 41: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/41.jpg)
Basic Properties of Expectation
MonotonicityIf X and Y are random variables such that X ≤ Y ”almost surely”,then E [X ] ≤ E [Y ].
LinearityE [aX + bY + c] = aE [X ] + bE [Y ] + c .
Jane Bae (Stanford) Probability and Statistics September 15, 2014 6 / 39
![Page 42: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/42.jpg)
Moments
The expected values of the powers of X are called the moments of X .
Mean: E [X ]
Variance: E [(X − E [X ])2] = E [X 2]− E [X ]2
(Standard Deviation =√
Variance)
There are higher order moments of both X and X − E [X ].
Jane Bae (Stanford) Probability and Statistics September 15, 2014 7 / 39
![Page 43: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/43.jpg)
Exercises
A number is chosen at random from the set S = −1, 0, 1. Let X bethe number chosen. Find the expected value, variance, and standarddeviation of X .
Let X and Y be independent random variables with uniform densityfunctions on [0, 1]. Find(a) E (|X − Y |).(b) E (max(X ,Y )).(c) E (min(X ,Y )).
Jane Bae (Stanford) Probability and Statistics September 15, 2014 8 / 39
![Page 44: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/44.jpg)
Covariance
The covariance between two jointly distributed real-valued randomvariables X and Y with finite second moments is defined as
σ(X ,Y ) = E [(X − E [X ])(Y − E [Y ])].
By using the linearity property of expectations
σ(X ,Y ) = E [XY ]− E [X ]E [Y ].
If σ(X ,Y ) = 0, the two random variables are uncorrelated.
Exercise: Show that independent random variables are uncorrelated.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 9 / 39
![Page 45: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/45.jpg)
Outline
1 Probability ConcetpsExpectation and MomentsConditional ExpectationMoment Generating Function, Characteristic Function
2 Limit Theorems
3 Statistics
Jane Bae (Stanford) Probability and Statistics September 15, 2014 10 / 39
![Page 46: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/46.jpg)
Conditional Expectation
For discrete random variables X and Y, the conditional expectation of Xgiven Y = y is
E [X |Y = y ] =∑x∈S
xpX |Y (x |y)dx =∑x∈S
xP(X = x ,Y = y)
P(Y = y).
Jane Bae (Stanford) Probability and Statistics September 15, 2014 11 / 39
![Page 47: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/47.jpg)
For continuous random variables X and Y, the conditional expectation ofX given Y = y is
E [X |Y = y ] =
∫ ∞−∞
xfX |Y (x |y)dx =
∫ ∞−∞
xfX ,Y (x , y)
fY (y).
Linearity and monotonicity also holds for conditional expectation.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 12 / 39
![Page 48: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/48.jpg)
Expectation Inequalities
Chebyshev’s Inequality:Let X be a random variable with finite expected value µ and finitenon-zero variance σ2. Then for any real number k > 0,
P(|X − µ| ≥ kσ) ≤ 1
k2.
Markov’s Inequality:If X is any nonnegative integrable random variable and a > 0, then
P(X ≥ a) ≤ E(X )
a.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 13 / 39
![Page 49: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/49.jpg)
Jensens Inequality:If X is a random variable and ϕ is a convex function, then
ϕ (E[X ]) ≤ E [ϕ(X )] .
Holder’s Inequality:If X and Y are random variables on Ω, and p, q > 1 with1/p + 1/q = 1,
E[|XY |
]≤(E[|X |p
])1/p (E[|Y |q])1/q.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 14 / 39
![Page 50: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/50.jpg)
Conditional Expectation Inequalities
Markov’s Inequality:If X and Y are any nonnegative integrable random variables anda > 0, then
P(X ≥ a|Y ) ≤ E[X |Y ]
a.
Jensens Inequality:If X and Y are random variables and ϕ is a convex function, then
ϕ (E[X |Y ]) ≤ E [ϕ(X )|Y ] .
Jane Bae (Stanford) Probability and Statistics September 15, 2014 15 / 39
![Page 51: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/51.jpg)
Holder’s Inequality:If X Y and Z are random variables on Ω, and p, q > 1 with1/p + 1/q = 1,
E[|XY ||Z
]≤(E[|X |p|Z
])1/p (E[|Y |q|Z])1/q.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 16 / 39
![Page 52: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/52.jpg)
Tower Property
Tower Property (Law of Iterated Expectation, Law of Total Expectation)
E [X ] = E [E [X |Y ]]
i.e.,E [X ] =
∑x∈S
E [X |Y = y ]P(Y = y)
If Y ∼ Unif (0, 1) and X ∼ Unif (Y , 1), what is E [X ]?
Jane Bae (Stanford) Probability and Statistics September 15, 2014 17 / 39
![Page 53: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/53.jpg)
Additional Properties of Conditional Expectation
E [Xg(Y )|Y ] = g(Y )E [X |Y ]
E [E [X |Y ,Z ]|Y ] = E [X |Y ]
E [X |Y ] = X , if X = g(Y ) for some g
E [h(X ,Y )|Y = y ] = E [h(X , y)]
E [X |Y ] = E [X ], if X and Y are independent
Jane Bae (Stanford) Probability and Statistics September 15, 2014 18 / 39
![Page 54: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/54.jpg)
Outline
1 Probability ConcetpsExpectation and MomentsConditional ExpectationMoment Generating Function, Characteristic Function
2 Limit Theorems
3 Statistics
Jane Bae (Stanford) Probability and Statistics September 15, 2014 19 / 39
![Page 55: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/55.jpg)
Moment Generating Function and Characteristic Function
Moment generating function and characteristic function chracterizes thedistribution of the random variable.
Moment Generating Function
MX (θ) = E [exp(θX )]
Characteristic Function
ΦX (θ) = E [exp(iθX )]
Jane Bae (Stanford) Probability and Statistics September 15, 2014 20 / 39
![Page 56: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/56.jpg)
Outline
1 Probability Concetps
2 Limit TheoremsModes of ConvergenceLaw of Large NumbersCentral Limit Theorem
3 Statistics
Jane Bae (Stanford) Probability and Statistics September 15, 2014 21 / 39
![Page 57: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/57.jpg)
Almost Sure Convergence
Let X1,X2, ... be a sequence of random variables. We say that Xn
converges almost surely to X∞ as n→∞ if
P(Xn → X∞ as n→∞) = 1
We use the notation Xna.s.→ X∞ to denote almost sure convergence, or
convergence with probability 1.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 22 / 39
![Page 58: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/58.jpg)
Convergence in Probability
Let X1,X2, ... be a sequence of random variables. We say that Xn
converges in probability to X∞ if for each ε > 0,
P(|Xn − X∞| > ε)→ 0
as n→∞. We use the notation Xnp→ X∞ to denote convergence in
probability.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 23 / 39
![Page 59: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/59.jpg)
Weak Convergence
Let X1,X2, ... be a sequence of random variables. We say that Xn
converges weakly to X∞ if
P(Xn ≤ x)→ P(X∞ ≤ x)
as n→∞ for each x at which P(X∞ ≤ x) is continuous. We use the
notation Xn ⇒ X∞ or XnD→ X∞ to denote weak convergence or
convergence in distribution.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 24 / 39
![Page 60: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/60.jpg)
Implecations
Almost Sure Convergence↓
Convergence in Probability↓
Weak Convergence
Jane Bae (Stanford) Probability and Statistics September 15, 2014 25 / 39
![Page 61: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/61.jpg)
Outline
1 Probability Concetps
2 Limit TheoremsModes of ConvergenceLaw of Large NumbersCentral Limit Theorem
3 Statistics
Jane Bae (Stanford) Probability and Statistics September 15, 2014 26 / 39
![Page 62: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/62.jpg)
Weak Law of Large Numbers
Theorem (Weak Law of Large Numbers)Suppose that X1,X2, ...is a sequence of i.i.d. r.v.s such that E [X1] <∞.Then,
1
n
n∑i=1
Xip→ E [X1]
as n→∞
Jane Bae (Stanford) Probability and Statistics September 15, 2014 27 / 39
![Page 63: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/63.jpg)
Strong Law of Large Numbers
Theorem (Strong Law of Large Numbers)Suppose that X1,X2, ... is a sequence of i.i.d. r.v.s such that E [X1] exists.Then,
1
n
n∑i=1
Xia.s.→ E [X1]
as n→∞
Jane Bae (Stanford) Probability and Statistics September 15, 2014 28 / 39
![Page 64: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/64.jpg)
Outline
1 Probability Concetps
2 Limit TheoremsModes of ConvergenceLaw of Large NumbersCentral Limit Theorem
3 Statistics
Jane Bae (Stanford) Probability and Statistics September 15, 2014 29 / 39
![Page 65: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/65.jpg)
Central Limit Theorem
Theorem (Central Limit Theorem)Suppose that X1,X2, ...,Xn are i.i.d. r.v.s with common finite variance σ2.Then, if Sn = X1 + · · ·+ Xn,
Sn − nE [X1]√n
⇒ σN (0, 1)
as n→∞. From here, we can deduce the following approximation:
1
nSn − E [X1]
D∼ 1√nN (0, 1)
Jane Bae (Stanford) Probability and Statistics September 15, 2014 30 / 39
![Page 66: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/66.jpg)
What is statistics?
Statistics is a mathematical body of science that pertains to thecollection, analysis, interpretation or explanation, and presentation ofdata, or as a branch of mathematics.
A statistic is random variable which is a function of the randomsample, but not a function of unknown parameters.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 31 / 39
![Page 67: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/67.jpg)
Outline
1 Probability Concetps
2 Limit Theorems
3 StatisticsEstimatorsEstimation Strategies
Jane Bae (Stanford) Probability and Statistics September 15, 2014 32 / 39
![Page 68: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/68.jpg)
Estimation
Estimation is making a best guess of an unknown parameter out ofsample data.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 33 / 39
![Page 69: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/69.jpg)
Estimators
An estimator is a statistic that is used to infer the value of an unknownparameter in a statistical model. (rule of estimation)
Examples:
Mean
Standard deviation
Jane Bae (Stanford) Probability and Statistics September 15, 2014 34 / 39
![Page 70: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/70.jpg)
Properties of Estimators
For a given sample x , the error of the estimator θ is defined as
e(x) = θ(x)− θ,
the mean squared error of θ is defined as
MSE(θ) = E[(θ(x)− θ)2].
The variance of θ is defined as
var(θ) = E[(θ − E(θ))2],
the bias of θ is defined as
B(θ) = E(θ)− θ.
Jane Bae (Stanford) Probability and Statistics September 15, 2014 35 / 39
![Page 71: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/71.jpg)
Properties of Estimators
An estimator is unbiased if and only if B(θ) = 0.
The MSE, variance and bias are related
MSE(θ) = var(θ) +(B(θ)
)2
Jane Bae (Stanford) Probability and Statistics September 15, 2014 36 / 39
![Page 72: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/72.jpg)
Confidence Interval
Consider the sample mean estimator θ = 1nSn. From CLT,
Sn − nE [X1]√n
⇒ σN (0, 1).
Rearranging terms, (this is not a rigorous argument)
1
nSnD∼ E [X1] +
1√nN (0, 1)
Jane Bae (Stanford) Probability and Statistics September 15, 2014 37 / 39
![Page 73: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/73.jpg)
Outline
1 Probability Concetps
2 Limit Theorems
3 StatisticsEstimatorsEstimation Strategies
Jane Bae (Stanford) Probability and Statistics September 15, 2014 38 / 39
![Page 74: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/74.jpg)
Maximum Likelihood Estimation
Finding most likely explanation:
θn = arg maxθ
f (x1, x2, ..., xn|θ) = f (x1|θ) · f (x2|θ) · · · f (xn|θ)
Gold Standard: Gueranteed to be consistent ( θmlep−→ θ0. )
Often computationally challenging
Jane Bae (Stanford) Probability and Statistics September 15, 2014 39 / 39
![Page 75: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/75.jpg)
Probability and Statistics
Jane Bae
Stanford University
September 16, 2014
Jane Bae (Stanford) Probability and Statistics September 16, 2014 1 / 32
![Page 76: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/76.jpg)
Overview
1 StatisticsHypothesis Testing
2 Markov ChainsMarkov ChainsAbsorbing Markov ChainsErgodic and Regular Markov Chains
3 Random WalksRandom WalksGambler’s Ruin
4 SimulationMonte CarloRare Event Simulation
5 Further InformationClasses at Stanford
Jane Bae (Stanford) Probability and Statistics September 16, 2014 2 / 32
![Page 77: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/77.jpg)
Outline
1 StatisticsHypothesis Testing
2 Markov Chains
3 Random Walks
4 Simulation
5 Further Information
Jane Bae (Stanford) Probability and Statistics September 16, 2014 3 / 32
![Page 78: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/78.jpg)
Sample Size
Note that as the degree of confidence increases, the interval must becomelarger.There is a way to improve both the degree of confidence and the precisionof the interval: by increasing the sample size. However, in the real world,increasing the sample size costs time and money.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 4 / 32
![Page 79: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/79.jpg)
Hypothesis Testing
Oftentimes we want to determine whether a claim is true or false. Such aclaim is called a hypothesis.
Null Hypothesis: A specific hypothesis to be tested in an experiment.
Alternative Hypoethesis: A specific hypothesis to be tested in anexperiment.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 5 / 32
![Page 80: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/80.jpg)
1. Formulate the null hypothesis H0 (commonly, that the observations arethe result of pure chance) and the alternative hypothesis Ha (commonly,that the observations show a real effect combined with a component ofchance variation).2. Identify a test statistic that can be used to assess the truth of the nullhypothesis.3. Compute the p-value, which is the probability that a test statistic atleast as significant as the one observed would be obtained assuming thatthe null hypothesis were true. The smaller the p-value, the stronger theevidence against the null hypothesis.4. Compare the p-value to an acceptable significance value α (sometimescalled an α value). If p ≤ α, that the observed effect is statisticallysignificant, the null hypothesis is ruled out, and the alternative hypothesisis valid.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 6 / 32
![Page 81: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/81.jpg)
Error in Hypothesis Testing
Type 1 Error: Reject the null hypothesis when it is true
Type 2 Error: Accept the null hypothesis when it is false
Jane Bae (Stanford) Probability and Statistics September 16, 2014 7 / 32
![Page 82: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/82.jpg)
Example
Suppose that ordinary aspirin has been found effective against headaches60 percent of the time, and that a drug company claims that its newaspirin with a special headache additive is more effective.
Null hypothesis: p = 0.6
Alternate hypothesis: p > 0.6,where p is the probability that the new aspirin is effective.
We give the aspirin to n people to take when they have a headache. Wewant to find a number m, called the critical value for our experiment, suchthat we reject the null hypothesis if at least m people are cured, andotherwise we accept it. How should we determine this critical value?
Jane Bae (Stanford) Probability and Statistics September 16, 2014 8 / 32
![Page 83: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/83.jpg)
Outline
1 Statistics
2 Markov ChainsMarkov ChainsAbsorbing Markov ChainsErgodic and Regular Markov Chains
3 Random Walks
4 Simulation
5 Further Information
Jane Bae (Stanford) Probability and Statistics September 16, 2014 9 / 32
![Page 84: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/84.jpg)
Markov Chains
A Markov chain is a sequence of random variables X1,X2,X3, ... with theMarkov property, namely that, given the present state, the future and paststates are independent.
P(Xn+1 = x | X1 = x1,X2 = x2, . . . ,Xn = xn) = P(Xn+1 = x | Xn = xn)
The possible values of Xi form a countable set S called the state space ofthe chain.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 10 / 32
![Page 85: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/85.jpg)
Markov chains are often described by a sequence of directed graphs, wherethe edges of graph n are labeled by the probabilities of going from onestate at time n to the other states at time n + 1, P(Xn+1 = x | Xn = xn).
Jane Bae (Stanford) Probability and Statistics September 16, 2014 11 / 32
![Page 86: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/86.jpg)
Specifying a Markov Chain
State Space: S = s1, s2, ...sr, the set of possible states.
Transition Probability: pij = P(Xn+1 = sj |Xn = si ).(Transition Matrix)
Jane Bae (Stanford) Probability and Statistics September 16, 2014 12 / 32
![Page 87: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/87.jpg)
Outline
1 Statistics
2 Markov ChainsMarkov ChainsAbsorbing Markov ChainsErgodic and Regular Markov Chains
3 Random Walks
4 Simulation
5 Further Information
Jane Bae (Stanford) Probability and Statistics September 16, 2014 13 / 32
![Page 88: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/88.jpg)
Absorbing Markov Chains
A state si of a Markov chain is called absorbing if it is impossible to leaveit (i.e., pii = 1). A Markov chain is absorbing if it has at least oneabsorbing state, and if from every state it is possible to go to an absorbingstate (not necessarily in one step).
probability of absorption
time to absorption
Jane Bae (Stanford) Probability and Statistics September 16, 2014 14 / 32
![Page 89: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/89.jpg)
Outline
1 Statistics
2 Markov ChainsMarkov ChainsAbsorbing Markov ChainsErgodic and Regular Markov Chains
3 Random Walks
4 Simulation
5 Further Information
Jane Bae (Stanford) Probability and Statistics September 16, 2014 15 / 32
![Page 90: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/90.jpg)
Ergodic Markov Chains
A Markov chain is ergodic if it is possible to go from every state toevery state (Also known as irreducible).
A Markov chain is regular if some power of the transition matrix hasonly positive elements.
Regular⇒ Ergodic
Jane Bae (Stanford) Probability and Statistics September 16, 2014 16 / 32
![Page 91: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/91.jpg)
The converse does not hold: Ergodic ; RegularEx) Let the transition matrix of a Markov chain be defined by
P =
(0 11 0
).
Jane Bae (Stanford) Probability and Statistics September 16, 2014 17 / 32
![Page 92: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/92.jpg)
Outline
1 Statistics
2 Markov Chains
3 Random WalksRandom WalksGambler’s Ruin
4 Simulation
5 Further Information
Jane Bae (Stanford) Probability and Statistics September 16, 2014 18 / 32
![Page 93: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/93.jpg)
Random Walks
Let Xk∞k=1 be a sequence of independent, identically distributed discreterandom variables. For each positive integer n, we let Sn denote the sumX1 + X2 + ...+ Xn. The sequence Sn∞n=1 is called a random walk.
Ex) If
Xi =
1, with probability 0.5
−1, with probability 0.5,
this is a symmetric random walk on a real line (R) with equal probabilityof moving left or right.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 19 / 32
![Page 94: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/94.jpg)
Probability of First Return
In the symmetric random walk process in R, what is the probability thatthe particle first returns to the origin after time 2m?
Jane Bae (Stanford) Probability and Statistics September 16, 2014 20 / 32
![Page 95: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/95.jpg)
Probability of Eventual Return
In the symmetric random walk process in Rm, what is the probability thatthe particle eventually returns to the origin?
For m = 1, 2 the probability of eventual return is 1. For other cases, itis strictly less than 1. (for m = 3, it is about 0.34)
Jane Bae (Stanford) Probability and Statistics September 16, 2014 21 / 32
![Page 96: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/96.jpg)
Outline
1 Statistics
2 Markov Chains
3 Random WalksRandom WalksGambler’s Ruin
4 Simulation
5 Further Information
Jane Bae (Stanford) Probability and Statistics September 16, 2014 22 / 32
![Page 97: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/97.jpg)
Gambler’s Ruin
Consider a nonsymmetric random walk on R.
Xi =
1, with probability p
−1, with probability q,
with p + q = 1.
A gambler starts with a stake of size s. He plays until his capital reachesthe value M or the value 0.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 23 / 32
![Page 98: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/98.jpg)
Outline
1 Statistics
2 Markov Chains
3 Random Walks
4 SimulationMonte CarloRare Event Simulation
5 Further Information
Jane Bae (Stanford) Probability and Statistics September 16, 2014 24 / 32
![Page 99: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/99.jpg)
Monte Carlo
Computational algorithms that rely on repeated random sampling tocompute their results.
Theoretical Bases
Law of Large Numbers guarantees the convergence
1
n
∑I(Xi∈A) → P(X1 ∈ A)
Central Limit Theorem
1
n
∑I(Xi∈A) − P(X1 ∈ A) ∼ σ√
nN (0, 1)
Jane Bae (Stanford) Probability and Statistics September 16, 2014 25 / 32
![Page 100: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/100.jpg)
Importance Sampling
We can express the expectation of a random variable as an expectation ofanother random variable.
eg.Two continuous random variable X and Y have density fX and fY suchthat fY (s) = 0 implies fX (s) = 0. Then,
E [g(X )] =
∫g(s)fX (s)ds =
∫g(s)
fX (s)
fY (s)fX (s)ds = E [g(Y )L(Y )]
where L(s) = fX (s)fY (s) is called a likelihood ratio.
Jane Bae (Stanford) Probability and Statistics September 16, 2014 26 / 32
![Page 101: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/101.jpg)
Outline
1 Statistics
2 Markov Chains
3 Random Walks
4 SimulationMonte CarloRare Event Simulation
5 Further Information
Jane Bae (Stanford) Probability and Statistics September 16, 2014 27 / 32
![Page 102: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/102.jpg)
Rare Events
Probability that a coin lands on its edge. How many flips do we need tosee at least one occurrence?
Jane Bae (Stanford) Probability and Statistics September 16, 2014 28 / 32
![Page 103: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/103.jpg)
Outline
1 Statistics
2 Markov Chains
3 Random Walks
4 Simulation
5 Further InformationClasses at Stanford
Jane Bae (Stanford) Probability and Statistics September 16, 2014 29 / 32
![Page 104: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/104.jpg)
Probability
Basic Probability: STATS 116
Stochastic Processes: STATS 215, 217, 218, 219
Theory of Probability: STATS 310 ABC
Jane Bae (Stanford) Probability and Statistics September 16, 2014 30 / 32
![Page 105: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/105.jpg)
Statistics
Intro to Statistics: STATS 200
Theory of Statistics: STATS 300 ABC
Jane Bae (Stanford) Probability and Statistics September 16, 2014 31 / 32
![Page 106: Jane Bae - Stanford Universitylanhuong/refresher/notes/prob... · 2015-05-31 · Overview 1 Probability Concepts Probability Space Combinatorics Random Variables and Distributions](https://reader034.vdocument.in/reader034/viewer/2022042411/5f2a344cb3a7e95d9d66fec5/html5/thumbnails/106.jpg)
Applications
Applied Statistics: STATS 191, 203, 208, 305, 315AB
Stochastic Systems: MS&E 121, 321
Stochastic Control: MS&E 322
Stochastic Simulation: MS&E 223, 323, STATS 362
Little bit of Everything: CME 308
Jane Bae (Stanford) Probability and Statistics September 16, 2014 32 / 32