I strongly believe the question you're asking is purely context specific.
For starters, I believe the "clear" button in your implementation is wrong because in a star-based rating system, a series of empty stars means the worst possible rating the control can accept. It's like grading something 0 in a 0-5 scale.
In the 2nd screenshot you provide it is Apple's way of requesting rating feedback when the user is about to delete an app. Funny thing, the usage of a "modal" UIAlertView (although they couldn't do it in another way) for that action is in direct violation of their own guidelines (iOS HIG), namely:
Ask for confirmation of user-initiated actions.
To get confirmation for an action the user initiated, even a potentially risky action such as deleting a contact, you should use an action sheet.
In the 3rd screenshot, the pseudo-modal view is used because the rating area (the red stars in the dimmed region) is clearly way smaller than the suggested minimum 44px area a human finger can coherently tap onto, so the developer decided to use a supplementary control that is displayed modally (to hierarchically draw attention) that features the same functionality, only in a size that's more comfortable to interact with.
In the following screenshot, you can see a view I designed for one of my client's projects. The rating controls (4 of them) are all included in the same hierarchical level and are designed so the user can clearly submit his feedback without interacting with any other controls. The labels prefixing the starred rows are "food", "services" and "ambience" in English and there's no need to explicitly state what 2 or 3 out of 5 stars means for food quality. It's like asking yourself, from 1 (worse) to 20 (best), what grade would I give to my girlfriend looks. A number makes sense to grade it. HOWEVER, in the last row, the user is asked to rate how "costly" is a Point Of Interest.... if you think this over, cost is something that needs to be expressed in quantity thus there are explanatory numbers in euro in parentheses right under the "rating" euro symbols.
Finally, rating more or less implicitly implies some sort of scale. For example, in YouTube videos it's either positive (thumbs up) or negative (thumbs down) rating. In EURO NCAP crash tests -where they rate a car's behavior in a crash- they use 1 to 5 stars. In the high-school I went they graded quizes, essays and tests from 0 (worst) to 20 (best) because there could be so many in-between states.
In my personal humble opinion, I'd assume a 5 star rating is sufficient for any mobile based rating system and you can display it anywhere you (or your designer) prefers, provided that each star is contained in it's own 44px "personal" tappable area.
All the best,
Thanos
