dataview.m

Sparse and possibly normalized representation of a dataset

A collection of one or more related @datasets, achieving a sparse (and possibly normalized) representation of a larger @dataset

Dataviews are compressed by providing a list of "key" variable names (analogous to key fields in a database) and a name for a new integer variable variable which will substitute for the key fields. This operation can be performed more than once.

Any dataset can be converted to a dataview by means of the dataview constructor, retrieved with dataview.full, and added to by means ofdataview.insert. The latter operation preserves the exising key lookups.

Motivations:

An ETL tool for relational databases
An updateable view (of sorts)
A "lite" database (one can combine several of these fellas)