Data analysis defines the following three methods.

Clustering Classification Regression
Clustering Klassifikation Regression
Classes unknown and similarities are
found from many data and formed into classes.
Classes have already been defined and
you only decide to which class a new object
should belong.
Identification of a trend line in the data sets and formation of relationships between data X and Y.

With 2D data X and Y this is almost obvious with the classes and trend lines, but face multidimensional data sets and try to find classes and trend lines or dependencies there! Only methods (neural networks) and tools (software tools) can help.