A Proposal for Some Standard Classes

We can facilitate the implentation of distributed methods for computing with data by identifying and defining (in IDL) some important data types that capture common concepts.

 o Variables: Sequences of values
IDL types such as sequence<double> or sequence<string> map into most systems and represent the common statistical concept of a sequence (vector) of values for a variable (observed or computed): numeric or character string values in the examples.

 o Data Table
a.k.a. data matrix, data frame. The concept is of a list of variables such as the above, forming the columns of a conceptual matrix or table.

Statisticians have tried to agree on a standard for this for at least thirty years. Are we now ready?

Many other classes may be useful. This proposal is not an attempt to fix the particular choices or their definition; that's for the community to decide. The main assertion is that such standards are helpful in writing distributed software, and that some, including the examples above, can be implemented well and easily.


How Standards Help Data Table Why are we doing this? Start
Last modified: Sun Jul 5 18:05:32 EDT 1998
© Lucent Technologies.