We have used Naive bayes algorithm and support vector machine to train the sample enron data having about 15000 each spam and non spam(ham) data

The data set can be downloaded from here http://www.aueb.gr/users/ion/data/enron-spam/ which is around 52.8 MB

The accuracy measured for Naive bayes implementation is about 99.08% for about 30% of random data from preprocessed data which was not used for training