Empirical Measures
From
Revision as of 15:28, 27 March 2009 by 209.237.84.181 (Talk)
Empirical analysis of a program is a factual enquiry carried out by simply recording what is observed or measured from actual run-time behavior of a program. Two of the more common empirical measures of program behavior are actual clock time and operations counting. Examples of operations counting: counting comparisons; counting assignments (memory writes); counting the number of times the inner most code in a loop is executed.
Asymtotic measures are "analytical" meaning that we look at the algorithm and deduce it's behavior. Empirical measures are situational observations; we measure real behavior for a specific situation. They are complimentary to each other; Empirical measures should validate our analytical view of program behavior.