p10:
part b typo: ...was within c of the actual value r
(should be b, right, like in part a?).

p11:
the problem needs more parts:
such as
what is the expected length of try
what is the expected number of tries
(for both parts).

p12:
paths are simple paths? (no cycles?)
if not then I don't see why the tree paths are not O(n).

p13-14-15: I don't like. 13 and 15 are similar to the pizza problem (counting
mistakes). 14 is repetition of exam 2 with the new parts being even more 
confusing and hard to grade.


General opinion:
I see very few problems on the untested material namely:
p1: induction
p2: state machines, graphs.
p3: series
p4: counting
p5: counting
p6: probability
p7: expectation
p8: counting, prob, little expectation
p9: prob
p10: sampling
p11: expectation
p12: graphs, orders
p13: counting
p14: counting, pigeonhole
p15: counting

Counting wins with 6/15!

We don't have a problem covering anything about variance, Markov, Chebychev,
Chernoff. 
3.3/15 problems have something relevant to the material not covered by the
first two exams. I thought Albert said that the final would be 50% biased
towards the later material.
Also I didn't see game-trees and structural induction. We didn't test
this either.

So, given these problems I would see something like:
p12 (maybe a couple of extra parts, this problem could be done in 2 minutes).
p1
p2
p3
p5
game-trees, pay-trees, etc
p6
p7
p8
p9
p10
p11 (if fixed)
some bound problem, like the spider in-class problem