Gutin and Yeo (2007, Algo Op Research) identified a class of optimisation problems (called anti-matroids) on which greedy attains the unique worst outcome.
user856 presented the construction on math.stackexchange in 2014 here.
Are there classification problems that correspond to anti-matroids? Specifically: I am trying to find an example of a classification problem in which a greedy decision tree attains the unique worst outcome.
