That's some real trash data.

Report is from 2017, the table is based on 2010-2014 ACS data. Issues at hand are multifold:
1. ACS data doesn't let you really filter by multiple criteria, all it is how many people in a vacuum tick each box for a given category. Trying to use this data to describe a specific subset of population is basically divination. Now, I did try to follow the chain of citations to try to see the methodology but I couldn't find their reference at all.
2. USCIS never released educational profiles of DACA recipients, all you've got are ages, sexes, countries of origin, and states of residence. Of note is that DACA doesn't even ask about education above high school.
3. Even if we were to presume that data from #1 is good, reweighting it by population delta is really a route to nowhere. If you're enrolled in college you'll stop being enrolled by one of two things happening - you graduate, or you drop out.
If we presume that 20% are currently enrolled in college after 3 years a significant chunk would've graduated or dropped out likely with higher graduation rate due to now being able to work and some states (like at the time Florida) allowing DACAers to get instate tuition rates.
On the flipside the new inflow until initial DACA applications got halted were primarily kids that aged into the program or the borderline cases who went back and got their GEDs. Though that trickle in likely wouldn't overpower the initial whiplash of people who suddenly could go to college doing so.
Weighing population trends requires the inflows and outflows to be fairly consistent.
4. I feel like using a data window where the program didn't even exist (pre-August 2012), is a really weird choice.

Now, who here has graduated after 2014? I have (2015). I'm not counted in the 5% number they pulled our their collective asses, according to them I was still in school 2 years after graduation.