Scribd Logo
Upload

Welcome to Scribd!

  • 16 views

    Bayesian Updating 2012

    Uploaded by

    Julie
    This document provides an introduction to Bayesian probability and Bayes' theorem. It defines key probability terms like the probability of a statement being true (p(S)), the probability of an event being true (p(E)), and the conditional probability of a statement being true given an event (p(S|E)). It then presents Bayes' theorem as p(S|E) = p(E|S)p(S)/p(E). The document explains these concepts using examples and concludes by applying Bayesian reasoning to calculate the probability that Romney will win the 2012 US presidential election given a potential event.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Bayesian Updating 2012

    Uploaded by

    Julie
    16 views5 pages
    This document provides an introduction to Bayesian probability and Bayes' theorem. It defines key probability terms like the probability of a statement being true (p(S)), the probability of an event being true (p(E)), and the conditional probability of a statement being true given an event (p(S|E)). It then presents Bayes' theorem as p(S|E) = p(E|S)p(S)/p(E). The document explains these concepts using examples and concludes by applying Bayesian reasoning to calculate the probability that Romney will win the 2012 US presidential election given a potential event.

    Original Description:

    slides on Bayesian Updating

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document provides an introduction to Bayesian probability and Bayes' theorem. It defines key probability terms like the probability of a statement being true (p(S)), the probability of an event being true (p(E)), and the conditional probability of a statement being true given an event (p(S|E)). It then presents Bayes' theorem as p(S|E) = p(E|S)p(S)/p(E). The document explains these concepts using examples and concludes by applying Bayesian reasoning to calculate the probability that Romney will win the 2012 US presidential election given a potential event.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    16 views5 pages

    Bayesian Updating 2012

    Uploaded by

    Julie
    This document provides an introduction to Bayesian probability and Bayes' theorem. It defines key probability terms like the probability of a statement being true (p(S)), the probability of an event being true (p(E)), and the conditional probability of a statement being true given an event (p(S|E)). It then presents Bayes' theorem as p(S|E) = p(E|S)p(S)/p(E). The document explains these concepts using examples and concludes by applying Bayesian reasoning to calculate the probability that Romney will win the 2012 US presidential election given a potential event.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 5

    Probability that a Statement Is True

    Probability that a statement is true: p(S). Often


    referred to as your prior belief (or just your
    prior).

    Bayesian Updating

    Examples:
    Matthew ate an orange today.
    Political Science 201b / Psychology 332b
    March 23, 2012

    Matthew intends to vote for Obama.


    Matthew will run for President.

    Probability that an Event Is True

    Conditional Probability: p(E|S)

    Probability that an event is true: p(E).

    Probability that an event is true conditional on S being


    true: p(E|S).

    Examples:
    All of the oranges are rotten.

    Examples:
    Probability that all of the oranges are rotten,
    conditional on Matthew having eaten an orange
    today.

    Matthew signs up with the Romney campaign.


    Matthew emigrates to Canada.

    Probability that Matthew signs up with the

    Romney campaign, conditional on Matthew


    intending to vote for Obama.
    3

    Conditional Probability: p(S|E)

    Bayes Theorem

    Probability that a statement is true conditional on E


    being true: p(S|E).

    Putting the four terms together, we get


    p (S|E) =

    Often referred to as your posterior belief or just


    your posterior. Its the main quantity of interest.

    p (E|S) p(S)
    ,
    p(E)

    which is Bayes Theorem.


    Examples:
    Probability that Matthew intends to vote for
    Obama, condition on Matthew joining the
    Romney campaign.

    Bayes Theorem tells us what our posterior should


    be, given the three other quantities.

    Why Should People Be Bayesian?

    Further Notes on Computation

     \ 
     \ 
    p S
    E =p E
    S

    (1)

    Suppose that S is either true or false, and E either


    happens or doesnt.

    T
    p (S E)
    p (S|E) =
    p(E)

    (2)

    Let S mean S is true and S mean not S or


    S is false.
    In this case, p(E) = p(E|S)p(S) + p(E|S)p(S).

    Conditional probability. . . Axiom 2 is conditional


    probability. It tells you the probability of one thing
    conditional on the probability of another thing. // If you
    accept the axioms, you are logically committed to
    Bayes.
    If you dont accept the axioms . . .

    This can help us . . .

    Further Notes on Computation

    Canonical Example: Disease Testing

    We now know p(E) = p(E|S)p(S) + p(E|S)p(S).

    p (S|E) =

    So Bayes Theorem is
    p (S|E) =

    p (E|S) p(S)
    p (E|S) p(S)
    =
    .
    p(E)
    p(E|S)p(S) + p(E|S)p(S)

    p (disease|pos. test) =

    p (E|S) p(S)
    p(E)

    p (pos. test|disease) p (disease)


    .
    p (positive test)

    This is useful because its sometimes easier to think


    about p(E) after we decompose it.
    9

    Disease Example, continued

    10

    Disease Example, continued

    The disease affects 1% of the population. So


    p (disease|pos. test) =
    p (disease|pos. test) =

    p (pos. test|disease) (.01)


    .
    p (positive test)

    p (pos. test|disease) (.01)


    .
    p (positive test)

    There is a test for the disease. It is 95% accurate:


    95% of people who have the disease test positive. So

    p (disease|pos. test) =

    11

    (.95)(.01)
    .
    p (positive test)
    12

    Disease Example, continued


    p (disease|pos. test) =

    Disease Example, continued

    (.95)(.01)
    p (pos. test)

    (.95)(.01)
    p (pos. test | dis.) p (dis.) + p (pos. test | ~dis.) p (~dis.)

    (.95)(.01)
    .
    (.95)(.01) + p (pos. test | ~disease) (.99)

    Now we only need to know the false-positive rate for


    the test, p (pos. test | no disease) .

    (.95)(.01)
    (.95)(.01) + p (pos. test | ~dis.) (.99)
    13

    Disease Example, continued


    =

    Disease Example, concluded


    Even after testing positive, your chance of carrying the
    disease is under 25%.

    (.95)(.01)
    (.95)(.01) + p (pos. test | ~disease) (.99)

    Many tests are better than this. But when you think
    about testing, think about false positives and chances
    of successful detection.

    Suppose that the false-positive rate is 3%. Then


    =

    14

    (.95)(.01)
    .24.
    (.95)(.01) + (.03)(.99)

    Note about base rates. In this example, we assumed


    that your prior was the base rate in the population.
    We didnt need to do that.

    15

    16

    Will Romney Win the 2012 Election?


    S is Romney will win.
    S is Romney will not win.
    p(S) = 2/5; p(S) = 3/5 (courtesy of inTrade).
    E is Romney giving up in MI before Election Day.
    p(E|S) = .2; p(E|S) = .5.

    p(S|E) =

    (.2)(2/5)
    21%.
    (.2)(2/5) + (.5)(3/5)
    17

    You might also like

    pFad - Phonifier reborn

    Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

    Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


    Alternative Proxies:

    Alternative Proxy

    pFad Proxy

    pFad v3 Proxy

    pFad v4 Proxy