Systems and methods for storing and accessing data using virtual data sets. Data sets may be removed from a data store and defined by algebraic relations between other data sets that are realized in the data store. A flag may be set to indicate that the data set is virtual. Criteria may be established for determining when a data set should be virtualized. For example, the criteria may be based on the size of the data set, the number of times it has been referenced and/or the frequency with which the data set has been accessed in the data store. A data set may also be optimized by partitioning the data set into subsets. The original data set may then be removed from the data store. An algebraic relation may be composed that defines the data set based on the subsets realized in the data store. The algebraic relation for the virtual data set may be used for optimizing access to other data sets even though the virtual data set is not realized.