If you want to check out another model, Lean Tossup has one for each province as well. I don't like their model as much as 338, plus their group are hyper-partisan and a bit brash and arrogant about it. As it stands right now though, their model is pretty in line with 338.
Popular Vote:
PC - 38%
LIB - 32%
GRN - 19%
PA - 7%
NDP - 4%
Seat Projection:
PC - 25 seats
LIB - 18 seats
GRN - 4 seats
PA - 2 seat
NDP - 0 seats
https://leantossup.ca/nb/
Differences:
- Miramichi is projected "leaning" a Liberal pickup by 338 by about 7%, projected by LTU a "tossup" PA hold by 0.2%
- Fredricton North is projected by 338 as a "toss up" PC pickup by 3.9%, projected by LTU a "safe" Green pickup by almost 12%
- Moncton East is projected by 338 as a "leaning" PC pickup by almost 7%, projected by LTU as "leaning" Liberal hold by 1%
- Moncton South is projected by 338 as a "tossup" PC pickup by 1%, projected by LTU as "likely" Liberal hold by 6%
- Victoria La Vallee is projected by 338 to be a "tossup" Liberal hold by 0.5%, projected by LTU to be a "likely" PC pickup by about 5%.
For what its worth, LTU did a good job projecting a final seat count for the 2019 NL election, but their district vote share projections were god awful. Not sure if their models have enough data or not. As an example from the district where I grew up:
Burin - Grand Bank LTU projection:
Liberal: 81.7%
PC: 18.3%
Safe Liberal hold
Election result:
Liberal: 51.6%
PC: 48.4%
They'll likely come close to their final seat projections, but there going to be very wrong about a bunch of districts but it evens out.