Why choosin' a conference instead of another?
Conference is a social event. Even in computer science where conferences really matter, attending and presenting a paper either in a talk, or as a poster is still a personal dissemination of results. With this prelude, I argue that you should choose also (if not primarily) on the basis of what kind of audience you expect to attend your presentation.
sometimes, your paper is a message to the community (incremental works or methodological remarks come to mind). In that case you want to target precisely the community your paper speaks to. So even if it would be a small workshop without proceedings, if your result matters to the small community, you should speak primarily to them.
most of the time, you have a standalone research result. Q: What are you after with it? A: impact = citations. Hence, go to a place attended by those guys who are likely to cite your work. Usually, that is correlated with the position of the venue on the Top-XYZ ladder = badge (see below).
sometimes, you want to get a badge = have a paper at this prestigious conference which will shine on your CV. Fair enough, there you have the way to decide.
Ideally, you should go for a place providing both 2 and 3. Honestly, I see little good from choosing a conference based solely on its focus. Doing so, you might end up at C-tier venue full of people who are interested in the location and the buffet.