The Enron-Spam dataset is a fantastic ressource collected by V. Metsis, I. Androutsopoulos and G. Paliouras and described in their publication "Spam Filtering with Naive Bayes - Which Naive Bayes?". The dataset contains a total of 17.171 spam and 16.545 non-spam ("ham") e-mail messages (33.716 e-mails total). The original dataset and documentation can be found here.
However, the original datasets is recorded in such a way, that every single mail is in a seperate txt-file, distributed over several directories. This can make reading in the data a bit cumbersome, especially for beginners. Since the data set is such an excellent ressource, I wanted to create a offer a single download of the data through a simple csv-file.
You probably only need the data file ("enron_spam_data.zip"). The python file ("build_data_file.py") contains the script used to construct the csv file (downloading the original raw data from the website, unpacking it, processing it and saving it into a compressed csv file).
Processing of the data is minimal: The dataset contains the following columns:
| Column | Explanation |
|---|---|
| Subject | The subject line of the e-mail |
| Message | The content of the e-mail. Can contain an empty string if the message had only a subject line and no body. In case of forwarded emails or replies, this also contains the original message with subject line, "from:", "to:", etc. |
| Spam/Ham | Has the values "spam" or "ham". Whether the message was categorized as a spam message or not. |
| Date | The date the e-mail arrived. Has a YYYY-MM-DD format. |