I would count bonus episodes, since we're tracking reader-retention, if a reader is engaged by your main story they would presumably see your bonus episode while scrolling through. Right?
Also, I think using the second newest episode is a lot more helpful, or otherwise using your newest episode but right before you upload your next episode (assuming you upload on a regular schedule) The purpose of that is so that the episodes you are calculating with have had their chance to "mature" and reach their peak amount of views during an upload-cycle.
I suppose the best way to figure that out would be to use the algorithm on series that you consider to be successful and compare yourself to the other series? Or perhaps instead compare yourself to a similar series that's been uploading for roughly the same amount of time/has the same amount of episodes.
Not sure, a hypothetical "bad series" let's say has
10 episodes. 50 subscribers.
100 views on its first episode, and 0 views on its most recent "mature" episode.
So there's a low views on first episode/sub ratio, and none of those subscribers have read the most recent episode.
(100-0)/10/50 = 0.5
A hypothetical "good series" lets say has
10 episodes, 50 subscribers.
100 views on its first episode, and 50 views on its most recent episode.
(100-50)/10/50 = 0.1
So I think lower numbers are better maybe?
EDIT: Also these two hypothetical series are pretty small potatoes. But the small numbers makes it easy to see how reader retention affects the final score. So a lower score = better reader retention I suppose?
One issue I'm seeing now is that this doesn't take into account your total viewcount and your average number of views across all episodes. Our hypothetical "bad series" could have had a perfect track-record before their 10th episode, and out Hypothetical good series could have had 0 views on episodes 2-9. This algorithm can't tell if that's happened. So it's probably not a perfect model and could be improved.
EDIT 2: A hypothetical "perfect" series, is one that has 100% reader retention, and consistently grows over time, and at that point your biggest factor in judging that series's success is the rate of growth right? Because if you have perfect reader retention your score will always be 0.
EDIT 3: This statistic crunching and evaluating might be off topic. Perhaps a new thread is in order?