Train a perception classifier to detect spam emails

Project structure:

Training script: perceptron_training.py, testing script: perceptron_testing.py, helper functions : perceptron_utils.py, bash overhead for training with all parameters included: perceptron_training.sh

Input parameters: -p : path to positive training data, -n : path to negative training data, -pp : path to positive test data, -nn : path to negative test data -m : model path

Call example of the training script: perceptron_training.py -p=enron1/ham/ -n=enron1/spam/ -pp=enron2/ham/ -nn=enron2/spam/ -m=model1

Both scripts also display training and testing errors

Dependencies: Python 2.7, nltk

Get the training an test sets from https://sites.google.com/site/rothbenj/teaching/repetitorium/perceptron.tgz?attredirects=0