Teaching‎ > ‎CS350 Spring 2009 TF‎ > ‎

2009-02-20

posted Feb 20, 2009, 11:14 AM by Site Admin Role   [ updated Feb 20, 2009, 11:56 AM ]
  • How to read Zα/2 from the table in lecture notes.
  • How many more samples do we need to get in order to reach a certain confidence interval? See more_samples.c attached.
  • Sampling a pseudo random number generator and see how many samples we need in order to reach the target error. See confidence.c attached.
    • Let zed = 1.96 for 95% confidence interval.
    • First use drand48(), uniform distribution [0.0, 1.0), S2 = 1/12.
      • E = 0.005, expected N = 12805
      • E = 0.004, expected N = 20008
      • E = 0.003, expected N = 35570
      • E = 0.002, expected N = 80033
    • Then use dexprand48() (hw1 problem 5), exponential distribution λ = 1.0, S2 = 1/(λ2) = 1.0
      • E = 0.005, expected N = 153664
    • In general, if you want to halve the error, you need to quadruple the number of samples.
ċ
Makefile
(0k)
Site Admin Role,
Feb 20, 2009, 11:33 AM
ċ
confidence.c
(1k)
Site Admin Role,
Feb 20, 2009, 11:33 AM
ċ
marko.c
(1k)
Site Admin Role,
Feb 24, 2009, 3:21 PM
ċ
more_samples.c
(1k)
Site Admin Role,
Feb 20, 2009, 11:33 AM
Comments