Numeric Operations on Instance Attributes

Various numeric manipulations can be performed to, for example,
examine the statistical distribution of the values or to compute new
instance attributes, such as the derivatives,normalizations, or the
'Critical Points' needed to form predicates,

Matrix:

The base information of the numeric analyses is a matrix of values.
The columns of the matrix are the individual attributes and the rows
of the matrix are the individual instances.  A matrix is formed by
choosing which instances and which attributes (within those instances)
should be used.

Distribution:

The basic distribution
of points can be examined and even further used as the basis of
the normalization of points.  Several flags are available for
normalization:

MeanOffSet:      The mean is used as the zero point
AverageOffSet:   The average is used as the zero point
NoScale:         No scaling is done
ZeroPreserving:  The zeros are preserved (The positive and negative
                 values are scales from zero).

Critical Points:

The heart of the creating fuzzy predicates automatically for numeric
values is the critical points.  These are a set of points and 
deviations from these points for each attribute.



