Not yet, there are a couple outliers that make the data set not really pass muster. I'm pretty sure with removal of the outliers that it would be fairly accurate, but I'd rather have the extra data to make them less significant.
On several generations I don't have enough data to provide meaningful results, although the trend line is pretty much as expected with later generations being smaller in size.
The biggest issue is currently to get to a 90% probability of weight, I end up with a range on the low end that I don't have data for. We'll just call it the Razi factor, and because he exists means I need more data to prove that he is far outside the norm.
I guess I could start imputing the missing data points, and get fairly accurate results although it's just way better to have real data.