#-------------------------------------------------------------------------------

MACHINE LEARNING - IMPLEMENTATION OF ID3 ALGORITHM FOR BINARY CLASSIFICATION

#-------------------------------------------------------------------------------

START DATE: 1/15/2014

END DATE: 2/3/2014

#-------------------------------------------------------------------------------

AUTHOR NAME : Pradeep Anatharaman

#-------------------------------------------------------------------------------

OPERATING SYSTEM USED: MICROSOFT WINDOWS X64 8.1

LANGUAGE USED : PYTHON , INTERPRETER VERSION : 2.7.6 \Pack available at:http://www.python.org/download/releases/2.7.6/

########SOURCE FILES###########

1. Id3BinaryClassification.py - Source file that contains the implementation of ID3 algorithm.

########FOLDERS###########

1. Data files - Contains files with training and testing data instances that were provided.

2. Learning curve - Contains files that were used to plot the learning curve

#########INSTRUCTIONS TO EXECUTE PROGRAM############################

1. The source file, training data file and testing data file should be placed in the same folder.

2. Open command/shell prompt and navigate to the folder where the above files are placed using "cd" command

3. Give command in the following format: Id3BinaryClassification.py <train_file_name> <test_file_name>

eg: id3.py train.dat test.dat

(refer "Screenshots.pdf" for more examples)

IMPORTANT : Ensure that the order of arguments do not change. i.e name of the file that contains training instances

should be given first followed by the name of the file that contains testing instances.

NOTE : If the environment variable is set appropriately for Python interpreter, then precede the above command # with the keyword "python". i.e python Id3BinaryClassification.py <train_file_name> <test_file_name>

4. If 2 input arguments are not given(more or less than 2), the program will throw an error and exit.

#########ASSUMPTIONS AND NOTES############################

1. In cases where decision cannot be made on the classification(Eg. when each attribute in the training instances

have same value), then randomly chosen class label is returned(in this program class label '1' is returned).

2. Attribute names have been hard coded to begin with string "attr" followed by its number. i.e attr1,attr2 and so on, # irrespective of the attribute names given in the train and test data files. The instances have been taken in as is.

######### END OF README ################################### ############################ ############################## ############################ ##############################