Lab 3 - Dealing with Uncertainty
This lab is worth 10 points.
Due: Friday January 25, 2008 at 5pm
The FORECAST model used in the book only covered one possible combination
of input: today is rain, rainfall is low, temperature is cold and sky is
overcast. There are other valid combinations that are not explored. For
this lab, we will look at another valid combination and see how the Bayesian
system differs from the certainty factor system.
Part 1: Bayesian system
The 6 rules for the full Bayesian version of FORECAST are listed on page 69 of
the second edition of the book. Using the same dialogue interface given on
pages 70 and 71, trace the probabilities for the following user responses:
What is the weather today?
=> dry
What is the temperature today?
=> warm
What is the cloud cover today?
=> overcast
Part 2: Certainty Factor system
The 6 rules for the certainty factor version of FORECAST are given on pages
80 and 81. Again, we will use the dialog from the book and the above
responses. However, we also have to prompt the user for the certainty in
each piece of evidence. Trace the probabilities for the following user
responses:
What is the weather today?
=> dry
What is the temperature today?
=> warm
To what degree do you believe the temperature is warm?
=> 0.8
What is the cloud cover today?
=> overcast
To what degree do you believe the sky is overcast?
=> 0.9
Write-Up
Turn in your traces for each system. Show all steps used to derive the
probabilities (such as first calculating the odds for the Bayesian model).
You may either email the write-up or turn in a hard copy.