Observations and Variables

We call a set of data derived from an object (experimental unit) an observation. Each object is measured according to various aspects, such as temperature, concentration of some constituents, frequency of occurrence of some phenomenon, etc. Each of these aspects is denoted as a variable.  By assembling all available data on all objects we can build a matrix - a table where the columns represent the variables and the rows represent the measured observations. Another common term for "variable" is feature.

As an example, we can take the results of a pre-election poll. The objects of the matrix are groups of individuals distinguished by age and sex. The variables are the percentages of their voting preferences, i.e. "Party Y" is a variable (feature), "age 55- (female)" is an observation (object).
 

Depending on the nature of the measured item and the measurement process the variables may be further classified into discrete and continuous variables.