Index of Tables


Overview of data

There are five columns of data at the top, three of which are subdivided into three sub-columns. The three sub columns are the values of the Best, Average, and Worst individual for every five generations. The five top-level columns are (in order):
  • Generation, which lists the generations that the data is taken from.
  • Fitness Measure, which is the value of the averages of the fitness measure for those generations
  • Complexity, which lists the average complexity of the resulting trees in those generations
  • Depth, which lists the tree depth of those generations
  • Variety of the population

    List of tables

    Average of results without complexity

    This table contains the results of the test that did not take the tree's complexity into account as a fitness measure (100 generations)

    Average of results with complexity

    This table contains the results of the test that used the tree's complexity as part of the fitness measure (100 generations)

    Adjusted average of results with complexity

    This table contains the results of the test that used the tree's complexity as part of the fitness measure, adjusted by subtracting the average complexity. The fitness measures here are equivalent to the measures of the test without complexity, and are used for comparision between the two tests. (100 generations)

    Difference between results with and without complexity

    This table contains the difference (test with complexity subtracted from the test without complexity) of the tests with and without complexity as a fitness measure. (100 generations)

    Absolute difference between results with and without complexity

    The table contains the difference (absolute value of previous tables) of the tests with and without complexity as a fitness measure. (100 generations)

    Average of results with win as factor

    This table contains the results of the test that used both Edgar's evaluation and the outcome of the game as a fitness measure. (200 generations)
    William Bauder
    wjb14@columbia.edu