Data sets include far more ancillary information than simple tables of numbers and the DataSet Object (DSO) was designed as a generic container that can hold all of this information. The standard DataSet object is similar to a Matlab structure but it is intended to standardize how data is organized and maintained. It was originally designed in 2000 by a consortium of chemical and engineering industry experts to contain the kinds of data that they worked with in real applications. It has been updated numerous times since then to add relevant functionality needed for multivariate analysis of physical science and engineering data.
The DataSet Object includes fields for sample and variable labels, class variables, time and wavelength axes, etc. It can contain standard discrete, time-series, or multi-dimensional image data. The basic data table is a single-type array and can contain any numerical or logical data types. One or more "class" sets can also be created to contain categorical information. Labels and classes can be used as an index into the DSO.
NOTE: This object does not require the Matlab Statistics Toolbox but is also not compatible with the Statistics Toolbox Data Array Object. Adjust your Matlab path accordingly if you plan to use both objects.
DataSet Object web page - http://www.eigenvector.com/software/dataset.htm
DataSet Object wiki documentation - http://wiki.eigenvector.com/index.php?title=DataSet_Object