Lab 10: Categorising with Decision Trees
- Here is the lense data from the UCI
benchmark, and the description
if you're interested.
- Make a decision tree the correctly categorises the first
four items. (The first four features are the inputs, and
the last is the category.)
- Allow the user to put in the item they're trying to categorise by
inputing the 4 features; output the category.
- If you'd like make a complete tree.
- Use the C4.5 algorithm to make a good decision tree.
- Implement the C4.5 algorithm to automatically learn the decision
- Split the data in half (say odds and evens). Learn
the tree from the first half and categorise the second half.
How well do you do?
- Now do the second fold.