Math 300 , Spring 2008, Day 4, Mon, Feb 4  Hit reload to get most current version

Repeat from Friday's  nonclass, plus some
Reading:
Review 1-2. Read 1-3.  Don't get wrapped up in "Double counting".  
Read ahead,  
Sets--OR (union), AND (intersection), NOT (complement), and probabilities thereof
Ash 1-4, pp 20-27  (postpone P(AorBorC) pp21 on first reading)
Math 251 did a little of this.4.2: M&M 5th ed: pp259-266 (not Independence), 4.5, pp 311-15 for and, or (notConditional)

HW:
Ash 1-2, p. 13:  5 (There's an easy way to do part b.  You get to choose which woman gets handed her coat first, second, etc.  The probabilities will be the same whichever woman you list first.  You can do it other ways but it's harder.)
7 (this is the most complicated one.  But still straightforward.  I had another sample space model, same answer: )
10 (Which events are Not (another event)?), 11

Ash 1-3,  p. 19: 1, 2
5  They mean "exactly" 3 W and 2R (how many "other" does that leave?)

4 (like example 4)
3 (harder? I think it's easier with an ordered list of 3.  After you've chosen the first person, who CAN'T you choose?) 6 Hint: let the people pick the seats.
10 Hints:  1) Will the probability be the same whether i = 1, 2, or 3...?. .  2) I suggest doing it with i = 1, for 2 boys, 3 girls, then building up, 4 boys, 6 girls.  Look for a general approach.

11 Optional--read answers.  I don't want to put bad ideas in your head.

- - - - - - - -
A)  Do Handout on Hypergeometric distribution.
B)  Use Siegrist's Virtual Laboratories  (  http://www.math.uah.edu/stat/) to simulate the Hypergeometric distribution:
Bottom of main page, Applets.  Finite Sampling Models> Ball and Urn Experiment. (Direct link to applets:
direct to applets (The i Applet instructions button gives good info about the applets in general. In the experiment applet, hover over the options for labels, or scroll down for text.)
Use 12 balls,  4 red, draw a sample of size 3. Single step a few times to see how it goes. Check your computation for the distribution in part A above. Reset.  Set Update 10, Stop 100.  Run.  See how close the bar graph of data matches (doesn't).  Run again and again (without resetting) until you like the match.  How many runs did you take? (scroll down the left data panel to see the last run)
Play with the pop. size, proportion of red, size of sample other values, look at the shape of the distribution in general.

C)  We find for Hypergeometric (not in Ash):  If you have N objects, and r are "defective", (or red)  , and you take a sample of n of the objects:  What is the probability of getting 0, 1, 2,...defectives in your sample?
Write down the general formula for x defectives.
Answer: hypergeometric,  N objects, r "defective",  a sample of n,
P(x defectives) = (rCx ×(N-r)C(n-x)) ÷ NCn.  In class we noted that this only works for x < n, x < r.  We also noted that when N=5, r=3, n=3, you could not have x =0 because there were not enough "good" objects.
  Find a general inequality involving x, with N, r, n, to reflect the possibility of not enough good objects.

D) Start a page
of "Named Probability Distributions."
  (Keep it separate...)
  The first is the Hypergeometric N objects, r "defective",  a sample of n,   P(x defectives)   Give the limitations (as from #C).
- - - - - - -
Old B) Work separately, HAND this one IN WEDNESDAY. FRIDAY  Using Example 4, pp. 15-16
a)  Show that the two probabilities of Method 1 and Method 2 (p.16)  are equal (set them equal, expand the "choose"'s, cancel numerators and denominators till the two sides match up.  You shouldn't have to actually multiply anything.)
b)  Ash chose to treat the sample space as "lineups" (Method 1:  36P7 in the denominator; finding the committee for the numerator and then multiplying it by 7! to get the possible lineups of the committee.  Method 2 starts with lineups in the top.)  You can also do the problem with a sample space of committees. Find the number of committees with 2 digits, 4 consonants, 1 vowel.  Find the count for the whole sample space. Show that the resulting probability equals the probability of Method 1.
= = = = = = = = = = = = = = = = = = = = = =
NOTES: "Combinatorics", continued.
Probability pattern is still: Find a sample space of equally likely outcomes; count them.  Then count the outcomes favorable to your event, divide by # in sample space.
Outcomes for many problems can be lineups, or committees. (Sometimes one is better, sometimes equally effective) Most common error: switching kinds between counting sample space and counting event.
Another Notation:  n(n-1)...(n-r+1) = [n]r 

- - - - - - - - - -

"Matching" problems: Can usually be made into a "slot" problem.  If you want to match two sets of individuals, make one set into a labeled "row" of slots (fixed order) and then distribute the other set.
Sometimes it will make a problem easier to "see" if you put labels on the things. 

E.g. Musical chairs.  5 kids and 3 chairs. How many ways can they sit down? (Oh, in the game the kids are in a fixed row,  harder.  Assume they're running across the room toward the chairs.)  Chairs 1__, 2__, 3__ .  Kids A, B, C, D, E.  First chair can get any of the 5, then next gets any of the remaining 4, 3rd gets any of the remaining 3.  5x4x3 .  3 kids and 5 chairs?  Let the kids be the slots.

- - - - - - - - - -
A standard "committee" pattern:  9 balls, 5 green and 4 red.  Draw 3 balls.  P(2 are red)=?.  "Hypergeometric"
Hypergeometric urn
   The 3 balls are a committee.  Number of Possible committees (sample space) = 9C3.
      How do we count the committees with exactly 2 red members?  Must have 2 red and 1 other (=green).  Think of them as the red subcommittee and the green subcommittee.  All-red subcommittee of 2 members, is drawn from 4 reds.  4C2.  All-green subcommittee of 1 member, is drawn from 5 greens.  5C1.  Any red subcommittee could be paired with any green subcommittee, so multiplication rule applies, there are (4C2 times 5C1) committees of 3 with exactly 2 reds.

Note nC1 = n.  Ash often writes "n" when "nC1" would illuminate the structure better.  Likewise, nC0 = 1, there is 1 way to choose an empty committee (= nCn, 1 committee of the whole)

This pattern can be extended to 12 balls: 5 green, 4 red, 3 blue. 
   Draw 4 balls.  Probability of 1Green, 2Red, (and therefore 1 blue):
    12C4 is total count for sample space.
    Top of the fraction: #(1G, 2R, 1B) :  5C1 (for the G) times 4C2 (for the R) times 3C1 (for the B)

Suppose you are in a room with 20 mosquitos and 5 carry malaria.  You get bit by 6.
   What is the prob.that  None are malarial?  1?  2?  3?  4?  5?  6?   (Hypergeometric)
General formula,
N objects, r "defective",  a sample of n,
P(x defectives) = (rCx ×(N-r)C(n-x)) ÷ NCn.   Note this only works for x < n, x < r.  (don't run out of balls, don'r run out of "defectives.")  (Other potential problem: run out of "good"="other" balls)

Supppse now that  3 of the non-malarial mosquitos carry dengue fever.  You get bit by 6.
  What is the prob. that you are bit by 3 malarial and 2 dengue mosquitos?  Only"clean" mosquitos?.

More complex combinations of these techniques (example 4 p. 15).  Again be careful: space is lineups or committees?



Sievers home  Math300-Sp08/Day8p4.htm  1pm 2/3/08
This page belongs to Sally Sievers who is solely responsible for its content. Please see our statement of responsibility.