- Oct 27, 2007
- 17,009
- 5
- 0
I'm studying for an upcoming exam and would love for an AI guru to help me with just one question. I'll try not to bore you with the details and just outline the problem.
We're given a table with 4 attributes, one of which is just a binary value for whether the tuple is acceptable. The others have a limited number of possible values . If you're familiar with decision trees and ID3 then you'll know what I'm talking about.
So the first part of the problem is to use ID3 to find the attribute to branch on first - that's cool, I just calculate the entropy and information gain on each attribute and find the one with the largest information gain.
The next part of the problem asks:
Using just this first attribute that you have determined, how well can you predict the class for (a, b, c)? Explain.
(where a, b and c are possible values from the three columns).
I'm not at all sure how to approach this. Help would be really, really appreciated.
Edit - the classes are just determined by that binary column, there are only two. By "acceptable" above I meant they belong to one particular class.
We're given a table with 4 attributes, one of which is just a binary value for whether the tuple is acceptable. The others have a limited number of possible values . If you're familiar with decision trees and ID3 then you'll know what I'm talking about.
So the first part of the problem is to use ID3 to find the attribute to branch on first - that's cool, I just calculate the entropy and information gain on each attribute and find the one with the largest information gain.
The next part of the problem asks:
Using just this first attribute that you have determined, how well can you predict the class for (a, b, c)? Explain.
(where a, b and c are possible values from the three columns).
I'm not at all sure how to approach this. Help would be really, really appreciated.
Edit - the classes are just determined by that binary column, there are only two. By "acceptable" above I meant they belong to one particular class.
Last edited:
Facebook
Twitter