Introduction to Probability and Sample Spaces

Size: px

Start display at page:

Download "Introduction to Probability and Sample Spaces"

Carmel Watkins
5 years ago
Views:

1 Introduction to Probability and Sample Spaces Prof. Tesler Math 186 Winter 2019 Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

2 Course overview Probability: Determine likelihood of events Roll a die. The probability of rolling 1 is 1/6. Descriptive statistics: Summarize data Mean, median, standard deviation,... Inferential statistics: Infer a conclusion/prediction from data Test a drug to see if it is safe and effective, and at what dose. Poll to predict the outcome of an election. Repeatedly flip a coin or roll a die to determine if it is fair. Bioinformatics We ll apply these to biological data, such as DN sequences and microarrays. Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

3 Related courses Math 183: Usually uses the same textbook and chapters as Math 186. Focuses on the examples in the book. The mathematical content is the same, but Math 186 has extra material for bioinformatics. Math 180BC plus 181BC: More in-depth: a year of probability and a year of statistics. CSE 103, Econ 120, ECE 109: One quarter intro to probability and statistics, specialized for other areas. Math 283: Graduate version of this course. Review of basic probability and statistics, with a lot more applications in bioinformatics. Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

4 2.2 Sample spaces Flip a coin 3 times. The possible outcomes are HHH HHT HTH HTT THH THT TTH TTT The sample space is the set of all possible outcomes: S = {HHH, HHT, HTH, HTT, THH, THT, TTH, TTT} The size of the sample space is N(S) = 8 Our book s notation S = 8 more common notation in other books We could count this by making a table: 2 choices for the first flip 2 choices for the second flip 2 choices for the third flip = 2 3 = 8 The number of strings x 1 x 2... x k or sequences (x 1, x 2,..., x k ) of length k with r choices for each entry is r k. Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

5 Rolling two dice Roll two six-sided dice, one red, one green: green red 1 (1, 1) (1, 2) (1, 3) (1, 4) (1, 5) (1, 6) 2 (2, 1) (2, 2) (2, 3) (2, 4) (2, 5) (2, 6) 3 (3, 1) (3, 2) (3, 3) (3, 4) (3, 5) (3, 6) 4 (4, 1) (4, 2) (4, 3) (4, 4) (4, 5) (4, 6) 5 (5, 1) (5, 2) (5, 3) (5, 4) (5, 5) (5, 6) 6 (6, 1) (6, 2) (6, 3) (6, 4) (6, 5) (6, 6) The sample space is S = {(1, 1), (1, 2),..., (6, 6)} = { (i, j) Z 2 : 1 i 6, 1 j 6 } where Z = integers N(S) = 6 2 = 36 Z 2 = ordered pairs of integers Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

6 DN sequences codon is a DN sequence of length 3, in the alphabet of nucleotides {, C, G, T }: S = {, C, G, T,..., TTT } How many codons are there? N(S) = 4 3 = 64 Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

7 continuous sample space Consider this disk (filled-in circle): y 3 3 C 2 x 3 Complications 3 S = { (x, y) R 2 : x 2 + y } The sample space is infinite and continuous. The choices of x and y are dependent. E.g.: at x = 0, we have 2 y 2; at x = 2, we have y = 0. Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

8 Events Flip a coin 3 times. The sample space is S = {HHH, HHT, HTH, HTT, THH, THT, TTH, TTT} n event is a subset of the sample space ( S): = First flip is heads = {HHH, HHT, HTH, HTT} B = Two flips are heads = {HHT, HTH, THH} C = Four flips are heads = (empty set or null set) We can combine these using set operations. For example, The first flip is heads or two flips are heads = { HHH, HHT, HTH, HTT } B = { HHT, HTH, THH } B = { HHH, HHT, HTH, HTT, THH } Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

9 Using set operations to form new events = First flip is heads = {HHH, HHT, HTH, HTT} B = Two flips are heads THT = {HHT, HTH, THH} TTH Union: ll elements that are in or in B TTT HHH HTT HHT HTH B = {HHH, HHT, HTH, HTT, THH} or B : The first flip is heads or two flips are heads This is inclusive or: one or both conditions are true. Intersection: ll elements that are in both and in B THH B = {HHT, HTH} and B : The first flip is heads and two flips are heads Complement: ll elements of the sample space not in c = {THT, TTH, TTT, THH} Not : The first flip is not heads Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26 B

10 Venn diagram and set sizes = {HHH, HHT, HTH, HTT} B = {HHT, HTH, THH} B = {HHH, HHT, HTH, HTT, THH} B = {HHT, HTH} THT TTH TTT HHH HTT Relation between sizes of union and intersection HHT HTH Notice that N( B) = N() + N(B) N( B) 5 = THH N() + N(B) counts everything in the union, but elements in the intersection are counted twice. Subtract N( B) to compensate. Size of complement N(B c ) = N(S) N(B) 5 = 8 3 B Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

11 lgebraic rules for set theory Commutative laws B = B B = B ssociative laws ( B) C = (B C) ( B) C = (B C) One may omit parentheses in B C or B C. But don t do that with a mix of and. Distributive laws (B C) = ( B) ( C) (B C) = ( B) ( C) These are like a(b + c) = ab + ac Complements De Morgan s laws c = S c = ( B) c = c B c ( B) c = c B c Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

12 Distributive laws Visualizing identities using Venn diagrams: (B C) = ( B) ( C) B C (B C) # B# # B# S# C# S# C# B C ( B) ( C) # B# # B# # B# S# C# S# C# S# C# Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

13 Mutually exclusive sets Two events are mutually exclusive if their intersection is. = First flip is heads = {HHH, HHT, HTH, HTT} B = Two flips are heads = {HHT, HTH, THH} C = One flip is heads = {HTT, THT, TTH} and B are not mutually exclusive, since B = {HHT, HTH}. B and C are mutually exclusive, since B C =. For mutually exclusive events, since N(B C) = 0, we get: N(B C) = N(B) + N(C) Events 1, 2,... are pairwise mutually exclusive when i j = for i j Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

14 2.3 Probability functions Historically, there have been several ways of defining probabilities. We ll start with Classical Probability: Classical probability Suppose the sample space has n outcomes (N(S) = n) and all of them are equally likely. Each outcome has a probability 1/n of occurring: P(s) = 1/n for each outcome s S n event S with m outcomes has probability m/n of occurring: P() = m n = N() N(S) Example: Rolling a pair of dice N(S) = n = 36 P(first die is 3) = P ( {(3, 1), (3, 2),..., (3, 6)} ) = 6 36 P(the sum is 8) = P ( {(2, 6), (3, 5), (4, 4), (5, 3), (6, 2)} ) = 5 36 Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

15 Classical probability Drawbacks What if outcomes are not equally likely? What if there are infinitely many outcomes? Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

16 Empirical probability Use long-term frequencies of different outcomes to estimate their probabilities. Flip a coin a lot of times. Use the fraction of times it comes up heads to estimate the probability of heads. 520 heads out of 1000 flips leads to estimating P(heads) = This estimate is only approximate because Due to random variation, the numerator will fluctuate. Precision is limited by the denominator flips can only estimate it to three decimals. More on this later in the course in Chapter 5.3. Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

17 Empirical probability E. coli has been sequenced: Position: Base: G C T T T T C T On the forwards strand: # s 1,142,136 P() = 1,142,136 4,639, # C s 1,179,433 P(C) # G s 1,176,775 P(G) # T s 1,140,877 P(T) Total 4,639,221 1 Sample space: set of positions S = {1, 2,..., } Event is the set of positions with nucleotide (similar for C, G, T). = {1, 9,...} C = {3, 8,...} G = {2,...} T = {4, 5, 6, 7, 10,...} Simplistic model: the sequence is generated from a biased 4-sided die with faces, C, G, T. Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

18 xiomatic probability definition of a probability function P based on events and the following axioms is the most useful. Each event S is assigned a probability that obeys these axioms: xioms for a finite sample space For any event S: P() 0 The total sample space has probability 1: P(S) = 1 For mutually exclusive events and B: dditional axiom for an infinite sample space P( B) = P() + P(B) If 1, 2,... (infinitely many) ( are pairwise mutually exclusive, then ) P i = P( i ) i=1 i=1 i=1 i = 1 2 is like notation, but for unions. Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

19 xiomatic probability additional properties dditional properties of the probability function follow from the axioms. P( c ) = 1 P() Example: P(die roll = 1) = 1 6 P(die roll 1) = = 5 6 Proof. and c are mutually exclusive, so P( c ) = P() + P( c ). lso, P( c ) = P(S) = 1. Thus, P( c ) = 1 P(). c P( ) = 0 Proof: P( ) = P(S c ) = 1 P(S) = 1 1 = 0 Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

20 xiomatic probability additional properties If B then P() P(B) Proof: Write B = ( c B). and c B are mutually exclusive, so P(B) = P() + P( c B). The first axiom gives P( c B) 0, so P(B) P(). B c U B P() 1 Proof: S so P() P(S) = 1. Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

21 xiomatic probability additional properties dditive Law If 1, 2,..., n are pairwise ( mutually exclusive, then n ) n P i = P( i ) n i=1 i=1 i=1 i = 1 2 n is like notation, but for unions. Prove for all integers n 1 using induction, based on: P ( ) ( 1 n ) n+1 = P(1 n ) + P( n+1 ) = ( P( 1 ) + + P( n ) ) + P( n+1 ). Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

22 xiomatic probability additional properties Induction proves the dditive Law for positive integers n = 1, 2,..., but not for n =, so we have to introduce an additional axiom for that: dditional axiom for an infinite sample space If 1, 2,... (infinitely many) are pairwise mutually exclusive, then ( ) P i = i=1 P( i ) i=1 Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

23 xiomatic probability additional properties c B U B B U U c U B B De Moivre s Law: P( B) = P() + P(B) P( B) Proof: P() = P( B) + P( B c ) P(B) = P( B) + P( c B) P() + P(B) = P( B) + ( P( B) + P( B c ) + P( c B) ) = P( B) + P( B) c U B c Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

24 xiomatic probability additional properties dditional properties of the probability function follow from the axioms: P( c ) = 1 P() P( ) = 0 If B then P() P(B) P() 1 dditive Law: If 1, 2,..., n are pairwise mutually exclusive ( i j = for all i j) then ( n ) P i = i=1 n P( i ) i=1 (The first three axioms only lead to a proof of this for finite n, but not n =, so n = has to be handled by an additional axiom.) De Moivre s Law: P( B) = P() + P(B) P( B) Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

25 Generalizing Sigma ( ) notation Sum over sets instead of over a range of consecutive integers Consider a biased n-sided die, with faces 1,..., n. The probability of face i is q i, with 0 q i 1, and q q n = 1: n q i = 1 i=1 For n = 8, the probability of an even number is: P(even number) = q s = q 2 + q 4 + q 6 + q 8 s {2,4,6,8} For a 26-sided die with faces,b,...,z, the total probability is P({, B,..., Z}) = q s = q + q B + + q Z = 1, s {,B,...,Z} and the probability of a vowel is P({, E, I, O, U}) = q s = q + q E + q I + q O + q U. s {,E,I,O,U} Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

26 Finite discrete sample space Each outcome has a probability between 0 and 1 and the probabilities add up to 1: 0 P(s) 1 for each s S P(s) = 1 For an event S, define P() = s P(s). Examples DN biased n-sided die (previous slide). s S For flipping a coin 3 times, P({HHT, HTH, THH}) = P(HHT) + P(HTH) + P(THH). Generate a random DN sequence by rolling a biased CGT-die. Deviations from what s expected from random rolls of a dice are used to detect structure in the DN sequence, like genes, codons, repeats, etc. Prof. Tesler Ch Intro to Probability Math 186 / Winter / 26

Lecture 1 : The Mathematical Theory of Probability

Lecture 1 : The Mathematical Theory of Probability 0/ 30 1. Introduction Today we will do 2.1 and 2.2. We will skip Chapter 1. We all have an intuitive notion of probability. Let s see. What is the probability