There's a much easier solution: the vast majority of legit auto-generated email will have a "List-Unsubscribe" MIME header. I filter based on that header into a 'Bulk' folder.
There are some exceptions, e.g. emails that go via the Drip service, but given limited number of automated email services you can probably catch 99% of emails with a few rules, rather than machine learning.
There are some exceptions, e.g. emails that go via the Drip service, but given limited number of automated email services you can probably catch 99% of emails with a few rules, rather than machine learning.