Homework 1 Implementation Details

• tictactoe - Each instance describes a tictactoe board. Each of the nine attributes describes one board position with either x, o, or b (for blank). The goal is to determine whether a given board configuration constitutes a win for the X player.
  • tictactoe.train - 200 instances
  • tictactoe.test - 50 instances

• mushroom and mushroom2 - Each instance describes a mushroom, using 22 attributes. The goal is to determine whether a given mushroom is poisonous or edible. Note: the mushroom test set has some attribute-values that do not appear in the training set (e.g. some of the test mushrooms have musty for their odor attribute, but none of the training mushrooms are musty). As stipulated in the assignment, you are not required to handle this (but there is a simple way to deal with it: when you are classifying an instance and reach a node where none of the possible attribute values match the instance, assign the most frequent class label occuring in the subtree rooted at that node). Therefore, you may use mushroom2 as input to your program instead. Please indicate clearly on your trace file which data set you used. Although they represent the same concept (edible vs. poisonous), the resulting decision trees are very different - as are their respective test-set performances.
  • mushroom.train - 300 instances
  • mushroom.test - 100 instances
  • mushroom2.train - 300 instances
  • mushroom2.test - 100 instances
• Note: You may find it helpful to create a smaller subset (10-20 instances) of the training data to work with initially while developing your code.

• connect4 - This data set is included here as an example of a harder data set for your ID3 code since there are three possible class values. (Your code does NOT have to run on it for either Homework 1 or Homework 2.) Each instance describes a connect4 board. There are 42 attributes named {a-g}{1-6}. The letter indicates the column of the board, and the number indicates the row (depth), i.e.:

     The board:  6 . . . . . . .
                 5 . . . . . . .
                 4 . . . . . . .
                 3 . . . . . . .
                 2 . . . . . . .
                 1 . . . . . . .
                   a b c d e f g
  

  • connect4.train - 200 instances
  • connect4.test - 100 instances