#1 Type Inference

It would be good to run a sampling algorithm on the attributes and determine the data-type. so as to create adequate structure when