Open MarcusKlik opened 4 years ago
synthetic
uses spline based numerical simulation based on 100 control points. So if we have less than 100 numerical levels (but many values), it's more efficient to just store these levels.
On the other hand, the number of levels might depend on the vector size. With the example above we could say, for example, that the number of levels is 1 percent of the number of values. And these levels have a distribution of their own...
two entirely different ways to approach the same vector...
A numeric vector can have a limited amount of levels that are replicated:
When do we regard the vector as a factor with numeric levels. How do we determine if these levels have a distribution of their own?