To expand slightly, the viability of dair->u-tilt is affected the following factors:
1) Weight class
2) Crouch-cancel
3) Location of characters after the dair
Lets assume you can turnaround u-tilt as easily as you can u-tilt, and that the turn-around animation is cancellable or non-existent if done correctly (I think this is the case). With this assumption we ALMOST eliminate the third factor, we must still consider spacings either side of the target at which u-tilt will connect but shine won't. Obviously, at these spacings you would try to u-tilt UNLESS your opponent is crouch-cancelling.
I think crouch cancel is the most/second-most important factor after spacing. It depends mostly on the player, which will be informed by their character (i.e. Peaches will usually try and CC downsmash if they can). It will also depend on the state of the mix up game. I.e. if you're been dair->shining or nair->shining they are unlikely to be CC'ing as they'll be looking to DI the shine (unless they're a particularly stubborn/scrubby Peach).
The last factor is weight class/percent. At 0% it's difficult (or theoretically impossible given frame perfect actionability) to get guaranteed follow ups off a shine on Peach, Puff, Luigi and anyone of similar weight/fall speed parameters. On these characters, ideally you would u-tilt always, UNLESS, once again they're crouch-cancelling.
On fast-fallers at low percents you pretty much always want to shine because while u-tilt can link to itself at low percents, it's heavily DI'able and after the second hit, or even first hit on very good DI, will not connect and puts you in lag.
On fast-fallers at mid-high percents is when u-tilt starts making sense as you can no longer link shine reliably and u-tilt often sets up for a bair or even an f-smash at particularly sweet percents and DIs.
