Testing my understanding here.....
You plan to measure a few things and predict a whole bunch more?
The best you could do is to get a relationship among the things you measure and predict one of them from the others.
But how does that translate to predicting variables you haven't even included in the measurements?
I think you should be asking this in a soil mechanics or soil nutrition or soil something anyway, forum.
(What do you even mean by "soil structure" anyhow?")