I have deployed several business applications that use ML using ruby and .NET. One was to keep an eye on rental construction equipment return forms and maintenance history to "tweak" when equipment gets service. Another is an application that monitors production line robotics and watches for many data variables that give clues to pending hardware failures. I also just finished some classifier code for a community site I manage that analyzes the post making sure it is applicable to the discussion and notifying a moderator if it is not.
As you can see there is a wide range of problems being managed in these examples. You can go on and on as "J" said below it's just statistics and having relevant data so the sky is the limit! Good luck!