Transcript document

Basic Probability
Statistics 515 Lecture 04
1
Importance of Probability
• Modeling randomness and measuring
uncertainty
• Describing the distributions of populations
• Obtaining descriptive measures of
populations
• Assessing uncertainty in the sampling
process
• Inference from sample to population
2
Basic Elements
• (Random) Experiments: “any activity!”
• Sample Space and Sample Points
– Venn Diagrams
– Tree diagrams
– discrete versus continuous
• Events: Simple and Compound
• Probability of Events and Interpretation
• Two obvious properties of probability
3
Some Examples of Experiments
• Toss a coin 3 times.
• A box has 100 balls, 1 red and 99 blues. Keep drawing a
ball, with replacement, until the red ball is obtained.
• Pick 2 people from the group: {Pete, Pauly, Mary, Magdy}
• Super Bowl!
–
–
–
–
Who wins?
Margin of Victory?
Length of game in minutes.
Yardage for M. Faulk.
• Credit card balance of a randomly chosen USC student.
• Genotype of progeny from heterozygote parents.
4
More Examples
• Earthquakes
– how many occurs in one year in the US of magnitude at least 2.0?
– Intensity of the next earthquake in CA
– Length of time between occurrences
•
•
•
•
Genotype of progeny from heterozygote parents
% return on an investment after one year
How long will the “Weakest Link” show last?
Many more!
5
Interpretation of Probability
• Frequentist interpretation
– “long-run” relative frequency
– weak law of large numbers
• Theoretical interpretation
– derived from a priori considerations
• Subjective interpretation
– ‘personal’ beliefs
6
Assigning Probabilities
For “equally likely” discrete sample spaces, which
are sample spaces in which the sample points all
have the same probability or chance of occurring,
N ( A)
P( A) 
N (S )
where N(A) = number of sample points in event A,
and N(S) = number of sample points in the sample
space.
7
Some Illustrations
• A fair die is to be rolled once. What are the
probabilities of each of the sample points?
• Suppose now however that two fair dice are
to be rolled once simultaneously. What will
be the sample space, and what are the
appropriate probabilities of each of the
sample points of the sample space?
8
Another Example
• Let us classify the students in this class by
Gender [Male (M) or Female (F)] and by
Class [Undergraduate (U) or Graduate (G)].
Suppose that the numbers per Gender-Class
combinations are as follows:
• N(M,U) = 16, N(M,G) = 4, N(F,U) = 22,
N(F,G) = 7.
• Experiment: Draw one student at random.
9
Some Questions
• What is the probability that a (male,
graduate) student is chosen?
• What is the probability that a male student
is chosen?
• What is the probability that the student is
either a male or a graduate student?
10
“Non-Equally” Likely Discrete Sample
Spaces
To assign probabilities to the sample points of
such sample spaces, the three conditions that
should be satisfied by the probabilities assigned
are:
• each probability must be at least zero;
• the sum of all the probabilities must equal 1;
• the probability assignment should reflect a priori
or subjective considerations, or long-run relative
frequency expectations.
11
An Illustration
• A biased die is to be rolled once. The die is
biased in such a way that the probability of
a face is proportional to the number of dots
of that face. What are the appropriate
probabilities of each of the faces?
12
Another Example
• A box has 100 balls, with 1 ball red, and 99
balls blue (“a lottery model”). Suppose the
experiment is to keep drawing a ball from
this box, with replacement, until a red ball
is obtained.
• What is the sample space and how do we
assign probabilities to the sample points?
13
Probability Operations
• Complements and Complementation Rule
• Union of Events and the Addition Rule for
Mutually Exclusive Events
• Intersection of Events
• Addition Rule for Non-Mutually Exclusive
Events
• DeMorgan’s Rules
14
Conditional Probability
• Idea behind the notion of conditional
probability
• Utility: probability updating
• Definition: P(B|A) = P(AB)/P(A)
• Reason for this definition
• Some Examples and Applications
15
Multiplication Rule
•
•
•
•
•
•
•
P(AB) = P(A)P(B|A)
P(ABC) = P(A)P(B|A)P(C|AB)
Applications of this concept
Revisiting the “lottery model”
Independent Events: P(AB) = P(A)P(B)
When do independent events arise?
Computing probabilities with independence
16
Updating Probability
•
•
•
•
•
•
Description of Setting
Concrete Situation
Theorem of Total Probabilities
Tree Diagram Demonstration
Bayes Theorem
Prior and Posterior Probabilities
17
An Example
• There are two coins in a box, with one coin being
an ordinary coin (has a Head and a Tail), while the
other coin being a two-headed coin. We draw one
of these coins at random, and without looking
which type of coin it is, we toss it once.
• What is the probability of getting a “Head” in this
toss?
• Suppose that the outcome of the toss is a “Head.”
What is the probability that we actually tossed the
two-headed coin?
18
Application On Medical Testing
• A test for HIV-infection has the following
characteristic: If the person is infected, the test
will have a positive result with probability of
0.9999; while if the person is not infected, the test
will return a positive outcome with probability of
0.001. In a certain population, 5 in every 10000
are HIV-infected. Suppose that one person from
this population is chosen at random and subjected
to this test. If the test turned out to be positive,
what is the probability that this person is HIVinfected?
19
First, A Test of Our Intuition!
• Without yet computing the probability, who
thinks that the probability that this person is
HIV-infected is above .90?
• Who thinks that the probability that this
person is HIV-infected is between 0.40 and
0.90?
• Who thinks that the probability is less than
0.40?
20