As the number of units approaches zero you need to knock on the door and survey the residents regarding their travel behaviour.

I would love to.  Unfortunately, its for a development that has not been built yet.

Volume 3 of 3 of the ITE TGM has some basic discussion with regards to the use of fitted curve equation vs. the ave. vehicle trip ends.  It basically says you have to use engineering judgement to select the appropriate method, which in simple terms means there is no straight-forward answer to your question.

It looks like there is a lot of data points in the (230) classification that are equivalent to your number of proposed units.  I would run both equations, and use the result which is closest to the other relevant data points on the data plot.

Accurately predicting traffic volumes for a future development is very difficult to do, so don't get too caught up in getting "exact" answers.

Okay I was being facetious. For 50 units, you will get closer to the mark surveying a nearby similar development. Car ownership rates, income level, demographics, etc. will play heavily on whether or not your numbers are reasonable. If I were a reviewer familiar with the neighborhood and you gave me straight numbers out of the ITE manual, I would raise an eyebrow. There would need to be justification for the numbers you choose and a demonstration that they are reasonable.