A couple of stats are not included in the weekly forecasts including yellow/red cards, penalty saves/misses, own goals and points deducted for conceding two or more goals. The first three events appear to be almost totally random save for the odd tendency by some particularly 'robust' players and thus don't seem material to our calculation (even if we could forecast them). Points conceded for conceding two or more goals is a potentially relevant number and will be included in the future once I find a way to properly estimate it. However, given that you should be looking at the better defensive teams each week anyway, it's unlikely that this total should play a major part in your weekly lineups/transfers.
Due to the nature of how the numbers are forecast, they shouldn't be compared to actuals over a very short period. The best way to use them is for ranking and context purposes on a weekly basis, only looking at exact player totals over a longer period.
The forecast data should be used alongside the other factors you would normally use when setting your team, rather than a one-stop answer to weekly picks. A 0.1 point difference is obviously slight so if your other research says to play Aguero (5.5) over Suarez (5.6) then go ahead and roll with the City man.