Comparing Decision Trees and Neural Networks

You may use:

for this assignment.

Generate or find two exemplar data sets with the following properties:

  1. A data set on which a neural network performs well but a decision tree learner performs statistically significantly worse.
  2. A data set on which a decision tree learner performs well but a neural net performs statistically significantly worse.
There are no lower or upper limits on the size of the data sets that you use. However, you must finish the assignment by the due date.

Graduate Students

Graduate students must generate or find two additional data sets with the following properties:
  1. A data set on which both neural nets and decision tree learners perform well. That is, there should not be a statistically significant difference in performance between the two.
  2. A data set on which both neural nets and decision tree learners perform poorly. That is, there should not be a statistically significant difference in performance between the two.
There is a lower limit of 100 on the number of exemplars in these data sets.

Good Luck

Turning it in

Please turn in
  1. Graphs showing the training and testing performance of your neural net on the data sets.
  2. A table showing the decision tree's performance on your data sets.
  3. Graphical representation of your decision trees
  4. A table showing the 10-fold cross-validated performance of your learning algorithms on the data sets supporting the statistical significance of your results.
  5. Links to your data sets.
Please let me know if you need extra disk space for these assignments.
Sushil Louis
Last modified: Wed Feb 22 9:45:52 PST 2006