The VARIOGRAM Procedure

Computational Resources

The fundamental computation of the VARIOGRAM procedure is binning: for each pair of observations in the input data set, a distance class and an angle class are determined and recorded. Let $N_ d$ denote the number of distance classes, $N_ a$ denote the number of angle classes, and $N_ v$ denote the number of VAR variables. The memory requirements for these operations are proportional to $N_ d \times N_ a \times N_ v$. This is typically small.

The CPU time required for the computations is proportional to the number of pairs of observations, or to $N^2 \times N_ v$, where N is the number of observations in the input data set.