stats 67 - lecture 7 & 8 video

Réussis tes devoirs et examens dès maintenant avec Quizwiz!

cumulative dsitribution function for binomial random variable X

F(x) =P(X<=x) =ΣP(X=x̄) =ΣC(n,x̄)*p^x̄*(1-p)^(n-x̄)

Properties of variance If c is a constant, then var(c) = __1__ If c is a constant, then var(cX) = __2__ If X and Y are independent random variables then var (X + Y) = __3__ As a result, if Xi's are independent random variables and Ci's are constants, then var(∑CiXi) = __4__

1 - 0 (what is the variance of 5 is like saying how much does it jump around -- which is 0) 2 - c^2 * var(X) (var(5x) = 25*var(c) -- the standard deviation then removes the power of 2) 3 - var(X) + var(Y) 4 - ∑Ci^2var(Xi)

The cumulative distribution function (cdf) of X is not just P(X = x) but __1__ We denote cdf as F(x) (where the pmf is f(x)). The cumulative distribution function is the __2__ Note that F(x) = P(X ≤ x) = 1 − P(X > x).

1 - P(X ≤ x) 2 - sum of several probability mass functions. The cumulative distribution function (cdf) of X is not just P(X = x) but P(X ≤ x). We denote cdf as F(x) (where the pmf is f(x)). The cumulative distribution function is the sum of several probability mass functions. Note that F(x) = P(X ≤ x) = 1 − P(X > x).

The expectation of X is denoted as E(X). E(X) = ∑xP(X = x) = ∑ xf(x) where ∑ lower bound is x∈Sx Can be viewed as __1__. For ease of notation, we let µ be the population expected value of X. That is to say E(X) = __2__

1 - averaging over all possible X values while weighting each possible value by its probability 2 - µ The expectation of X is denoted as E(X). E(X) = ∑xP(X = x) = ∑ xf(x) where ∑ lower bound is x∈Sx Can be viewed as averaging over all possible X values while weighting each possible value by its probability

Bernoulli distribution handles ___1___ Set Sx = {0, 1}. Let P(X = 0) = 1 − p and P(X = 1) = p, where p is some number between 0 and 1 (a probability). Then we say X follows a Bernoulli distribution with parameter p. Note that f (x) = P(X = x) = p^x (1 − p)^(1−x) for x = {0, 1}. -- important -- Because of the fact that P(X = 0) = 1 − p and P(X = 1) = p, we get E(X) = 0 ∗ (1 − p) + 1 ∗ p = p. And thus, var(X) = E(X^2 ) − (E(X))^2 = p − p^2 = p(1 − p).

1 - binary outcomes. P(X = 0) = 1 − p and P(X = 1) = p means that it can be an unfair outcome, 0 shows up with probability 1 - p and 1 shows up with probability p, so it depends on what p is p^x (1 − p)^(1−x) plug in x=1 you get p

expectation of x general formula and expectation of x as a binomial distribution

1 - general: Σxf(x) 2 - x*C(n,k)a^k*b^n-k

In general, expectations of X are noted as the __1__ of X. Thus E(X^k ) is noted as the k-th __2__ of X, for k = 1, 2, 3, 4, ....

1 - moments 2 - moment

Each value of x(x=0,1,2,...n) has C(n,x) ways of occurring. Each of these events has probability p^x(1-p)^n-x ways of occurring (x many successes and n-x nonsuccesses) We can now create the ___1___ of X using functions. f(x) = P(X=x)=C(n,x)p^x(1-p)^n-x for x=0,1,2,...,n You can verify that f(x) is a valid probability mass function because f(x) = P(X=x)=C(n,x)p^x(1-p)^n-x >= 0, f(x) is only defined for 0<=x<=n and because Σf(x)=1 which can be proven with __2__ n and p are paramters -> n is known, p is unknown

1 - probability mass function 2 - the binomial theorem, which states that (a+b)^n = ΣC(n,k)a^k*b^n-k, setting a=p and b=1-p

f (x) is the __1__ of a random variable X. The input is __2__, which is a specified value of X from its support. The output is __3__ the probability that X is equal to what we specified, x. Thus f (x) = P(X = x). Since it is a probability, 0 ≤ f (x) ≤ 1. And all the outputs must sum to 1.

1 - probability mass function 2 - x 3 - P(X = x) f (x) is the probability mass function of a random variable X. The input is x, which is a specified value of X from its support. The output is P(X = x), the probability that X is equal to what we specified, x. Thus f (x) = P(X = x). Since it is a probability, 0 ≤ f (x) ≤ 1.

binomial distribution We say X follows a binomial dsitribution with parameters n and p. X is the number of ___1___ in n many independent trials Each trial has the probability ___2___ of having a success, and so it has the probability of __3__ of not succeeding This is like counting the number of heads that occur in n many independent flips (this specific case of flipping a coin to heads has p=0.5)

1 - successes, yesses, or heads 2 - p 3 - 1-p

Can derive the expectation and variance in an easier fashion when we construct the binomial random variable as the sum of n many independent and identical Bernoulli random variables. The binomial random variable X can be seen as the __1__ That is to say, if Zi are independent and identical Bernoulli random variables with p, then X=ΣZi E(X)=E(ΣZi)= ΣE(Zi)=Σp=np And variance: var(X) = var(ΣZi) = Σvar(Zi) = Σp(1-p)=np(1-p) What is f(x) and E(x)?

1 - sum of n independent and identical Bernoulli random variables 2 - f(x) = P(X=x) = C(n,x)p^x(1-p)^n-x for x=0,1,2,....,n 3 - E(x) =np and var(X) = np(1-p)

The the variance of x is defined to be __1__ var(X) = __2__ We denote the variance of X as __3__ Since it is the expected value of a squared random variable, σ^2 > 0. σ = sqrt(var(X)) is known as the __4__

1 - the average squared deviation from the mean/expected value/average (which is E(X)). 2 - var(X) =E[(X − µ)^2 ] =E[(X − E(X))^2 ] = E(X^2)-[E(X)]^2 3 - σ^2 4 - standard deviation of X

Note with the properties of variance, it is straightforwards to compute things of the form var(5 + 4X) once we have computed var(X). var(5 + 4X)=4^2*var(X). In general, for constants a and b: ___1___

1 - var(a+bX)=b^2 * var(x).

Let h(X) be a function of X, such as h(X) = 1/X E(h(X)) __1__ h(E(X)). Thus E(1/X) != 1/E(x) As a result (remember the formula for variance depends on expectations), var(h(X)) __2__ h(var(X))

1.) !=, does not equal 2.) !=, does not equal

Can now create the probability mass function of X, f (x), to define its probability distribution. What formula would you use to find f(x) = P(X=x) for flipping a coin n times and keeping track of the heads.

C(n,x)*(1/2)^n

variance of X

E(X^2)-[E(X)]^2

Say X is a random variable that follows a binomial distribution with n and p (for example n = 100 and p = 1 / 4). What is the expectation of x and the variance of x? What is P(x=5)? P(x=55)?

E(x = np=100*1/4) var(x)=np(1-p)=100*1/4*3/4 P(x=5)=C(100,5)*p^5*(1-p)^100-5 =(100!/(5!95!))*(1/4)^5*(3/4)^95 P(X=55)=C(100,55)*p^55*(1-p)^(100-55)=(100!/(55!55!))*(1/4)^55*(3/4)^45

Say n=5 and outcome is Y (yes or 1) and N (no or 0). Elements in this sample space are the outcome of 5 trials, in order. Now say x=2. Thus x=2 and n-x=3. An element from the sample space that has x=2 is YYNNN or NYNYN. P(YYNNN): P(Y and Y and N and N and N) =P(Y)P(Y)P(N)P(N)P(N)=pp(1-p)(1-p)(1-p)=p 2 (1 − p) 3 . Also P(NYNYN): P(N and Y and N and Y and N) = ??

P(N)P(Y)P(N)P(Y)P(N) =(1-p)p(1-p)p(1-p) =p^2(1 − p)^3

Say X is a random variable that follows a binomial distribution with n and p (for example n = 100 and p = 1 4 ). Compute the probability that we see less than 99 successes out of the 100 attempts. This is to say P(X < 99) = P(X ≤ 98). There are two ways to do this.

Remember we can use the complement rule. Say event A is X < 99 and so A^c is X ≥ 99. With a binomial random variable X, then F(98) = P(X ≤ 98) = ΣP(X=x̄) =Σ(100, x̄)*(1/4)^x̄*(3/4)^n-x̄ where x̄ <= 98 = C(100, x̄)(1/4)^ x̄(3/4)^n- x̄ So we compute this for 99 pieces C(100,0)(1/4)^0(3/4)^100-0 + C(100,1)(1/4)^1(3.4)^100-97 + ....+ C(100, 97)(1/4)^97(3/4)^100-97+C(100,98)(1/4)^98(3/4)^100-98 *BUT using the complement rule*: P(X ≤ 98) = 1 − P(X > 98) = 1 − P(X ≥ 99) To compute this, will have 2 pieces to calculate. P(X ≥ 99) = P(X = 99) + P(X = 100) 1 - [C(100,99)(1/4)^99(3/4)^100-99 + C(100,100)(1/4)^100(3/4)^100-100] = 1 - C(100,99)(1/4)^99(3/4)^100-99 + C(100,100)(1/4)^100(3/4)^100-100

Now extend the flipping a coin 3 times example to flipping a coin a large number of times, say 100 times. Sx = {0, 1, 2, 3, ..., 99, 100} The sample space will contain 2^100 elements, each a 100 character long string of HHHHH..... or HTHHTHT..... For each value of x in Sx, can think of it as arranging x many heads among 100 spots The probability of each event in the sample space is (1/2)^100 Each value of x will need to determine how many of the 2^100 events in the sample space will result in that x value. Each value of x will need to compute the number of ways of selecting x many objects from a total of n = 100 of them to pick from. This is the C(n, k) = n!/k!(n-k)! formula from before. Each value of x = 0,1,2,...,100 will have the probability computes as multiplying the number of ways x can occur with (1/2)^100 Sx = {0,1,2,3,...,99,100} What is P(X=20)

Take x = 20 Would need the number of elements in S (the sample space) that have 20 H's arranged amoing the 100 spots (and the other 80 by T's) Using the formula, we get there are 100!/20!(100-20)! total ways of having 20 heads show up in 100 flips. The probability of each event (20 H's and 80 T's) is (1/2)^100 Then P(x=20) = 100!/20!(100-20)!*(1/2)^100

For example, P(10 < X ≤ 14) = P(11 ≤ X ≤ 14) means

This is to say the probability of X being less than or equal to 14 and greater than 10. P(11 ≤ X ≤ 14) = P(X = 11) + P(X = 12) + P(X = 13) + P(X = 14)

Suppose you play a game where you can either win a dollar or lose a dollar. It's as equally likely that you'll win a dollar as it is that you'll lose a dollar. Say Sx= {−1, 1} where P(X = 1) = 0.5 and P(X = −1) = 0.5 What is the variance of X? What is the standard deviation?

We know variance of X is E(X^2)-[E(X)]^2 We first find E(X) = -1 *.5 + 1*.5 = 0 E(X^2) = -1^2 *.5 + 1^2*.5 = 1 We plug into our variance formula var(X) = 1 - 0^2 = 1 The standard deviation is sqrt(1) = 1

Suppose you are in a scenario where you either lose 1,000 dollars or win 1,000 dollars. Each is equally likely. Say Sx = {−1000, 1000} where P(X = 1000) = 0.5 and P(X = −1000) = 0.5 What is the variance of X?

We know variance of X is E(X^2)-[E(X)]^2 We first find E(X) = -1000 * .5 + 1000 * .5 = 0 E(X^2) = -1000^2 * .5 + 1000^2 * .5 = 1000^2 We plug into our variance formula var(X) = 1000^2 - 0^2 = 1000^2 The standard deviation is sqrt(1000^2) = 1000

Example is rolling a die once. Our pmf table will have 1/6 for x = 1,2,3...6 and f(X=x). What is the variance of X?

We know variance of X is E(X^2)-[E(X)]^2 We first find E(X) = 1 * 1/6 + 2* 1/6 + 3 * 1/6 + 4 * 1/6 + 5 * 1/6 + 6 * 1/6 = 3.5 E(X^2) = 1^2 * 1/6 + 2^2* 1/6 + 3^2 * 1/6 + 4^2 * 1/6 + 5^2 * 1/6 + 6^2 * 1/6 = ~ 15.1 E(X^2)-[E(X)]^2 = 15.1 - 3.5^2 = ~2.91

Suppose we are using the example of flipping a coin 100 times and keeping track of when it lands on heads. Sx={0,1,2,3,...99,100} Can do this for all values of x =0,1,2,...100 What is P(X=0)?

We use the formula of (choose/combinartoric) * probability of the event C(100,0) = 1 total way of having 0 heads show up in 100 flips The probability of this event (100 tails in a row, TTTT....) is (1/2)^100. Then P(X=0) = 1*(1/2)^100

Can now create the probability mass function of X, f (x), to define its probability distribution of a binomial random variable. Counting the number of events is the same, but the probability of each event in the sample space (arrange x many success among n many trials, for x = 0, 1, 2, ..., n). Each event in the sample space will have probability involving p and (1 − p). Specifically, each event that has x many success among the n many trials is p^x * (1 − p)^n−x . This is the product of n many independent and identically distributed Bernoulli random variables. Why is each element in the sample space of the form p^x * (1 − p)^n−x?

Where x is the number of yes' (or 1's or success') and n is the number of no's. Note that all n many trials are independent, where the yes trials have probability p and the no trials have probability (1-p). To get a specific element in the sample space (n trials, a combination of yes' and no's), there are x many yes' and n-x many no's. Thus the probability of a specific element in the sample space will be p^x * (1 − p)^n−x , where the p*x are from the yes' and the (1 − p)^n−x are from the no's.

Say X is a random variable that follows a binomial distribution with n and p (for example n = 100 and p = 1/4 ). Compute the probability that we see less than 4 success's out of the 100 attempts. This is to say P(X < 4) = P(X ≤ 3) (X = 0, 1, 2, or 3)

c(100,0)*(1/4)^0*(3/4)^100-0 + c(100,1)*(1/4)^1*(3/4)^100-1 + c(100,2)*(1/4)^2*(3/4)^100-2 + c(100,3)*(1/4)^3*(3/4)^100-3

Flipping coin 3 times example: example of flipping a coin 3 times and the event of landing on heads for each flip. The support of x, Sx = {0,1,2,3} But the sample space had 8 elements, S = {HHH, HHT, HTH,THH, HTT,THT,TTH,TTT}. For each value of x in Sx , can think of it as arranging x many heads among 3 spots. When x = 1 for example, it is the same as placing a single H among the 3 spots, such as HTT, THT, and TTH. C(3,1) which is 3. So the probability of x = 1 is 3*(1/2)^3 When x = 2 it was placing two H's among 3 spots. HHT, HTH, THH, C(3,2) which is 3. So the probability of x = 2 is 3*(1/2)^3. This is (1/2)^3+(1/2)^3+(1/2)^3 since we're using the mutually exclusive rule. Can see it as selecting x many objects out of 3. How many of the spots will be given an H designation. We have 3 spots, and need to choose x many of them (for x = 0, 1, 2, 3).

example


Ensembles d'études connexes

Business Analytics 2 Professor Winkofsky Final Exam

View Set

Chapter 72: Caring for Clients with Dementia and Thought Disorders

View Set

She Sells Cell Phones Unit Reviews

View Set

Input, Storage and Output Devices

View Set

Leveraged Finance Interview Technical Questions

View Set