Performance Analysis*

Note that while all three versions of our program work, Version 3 is the best because it is more efficient, and more readable. (Increased efficiency does not always go hand in hand with increased readability). The table below summarizes the relative numbers of comparison made by each of the versions.

  ----------------- ----------------- ----------------- -----------------
                    Number of                           
                    comparison &                        
                    logical                             
                    operations                          

                    Minimum           Average           Maximum

  Version 1         12 if/elifs @ 3 = 12 if/elifs @ 3 = 12 if/elifs @ 3 =
                    36                36                36

  Version 2         3                 (12 if/elifs @ 3  12 if/elifs @ 3 =
                                      )/2 = 18          36

  Version 3         3                 ((1 if @ 3) +\    ((1 if @ 3) + (11
                                      (11 if/elifs @ 1  if/ellifs @ 1 ))
                                      ))/2 = 7          = 14
  ----------------- ----------------- ----------------- -----------------

Notes:

1. We might be able to make the program even faster by by checking for the most likely grades first (e.g. C's first, then B's, then A's, then F's). This would minimize the number of tests that need to be done, but the resulting loss of readability might not be worth the performance gain.

2. The table overstates the numbers of comparisons that will actually be made in some cases. The reason is that Python "short circuits" test expressions when it can. More on this later.

• This is a theoretical performance analysis in which we are estimating the expected performance. We will soon also be doing empirical or experimental performance analyses in which we measure actual performance rather than trusting to fallible predictions.