Raw Data feeds are tricky and require a lot of rules to be applied to use properly.. While I don't work directly with the raw data feeds, I work closely with our data lake team and help them to understand the raw data and process it... so I don't have formulas per se, but I can help with the fields of interest to you.
First, you should make sure that you are using the exclude_hit column to remove data that should not be processed.
Second, you might want to look at only page views (these should be the rows where you have pagename and page_url (or at least where you are looking at Single Page Views.. you would want to see all hits if you are trying to match Adobe's bounce rate definition - more below)
Now visit_page_num should indicate the page number in the visit, so items with 1 should be your entry pages. Each page view should increment this value per user, per visit.
You will need to identify users either by using mcvisid or visid_high and visid_low (it all depends on what data you have). By correlating your visitor identifier with visit_num (which should increment since the beginning of time for that user identification), you should be able to find the visits that have one page view, or multiple page views (using the above visit_page_num).
- Adobe calculates "Bounces" as a single hit per visit (so in that case, you wouldn't want to filter out the non-page view hits - if the user triggered any clicks on that one page, they won't be counted as a bounce - assuming you want to use that definition of "bounce" - we prefer to track bounces as single page view per visit but that is a personal choice that you will have to determine for yourself).
- Entries should be the visit_page_num = 1 for each visit per user
- Single Page Views should be the users that only have 1 page view (hits with pagename and page_url) within the same visit.
Hopefully this will get you started, good luck.
