Analysis of robustness about decision tree induced by insensible attribute

There have been many heuristics to induce decision tree. The traditional ones include ID3 and min-ambiguity etc. We have proposed one new heuristic for generalizing decision tree and done some experiments to have proved its merits. We do more experiments to support that the insensible method is more robust than others and analyze the reason of robustness about the insensible method.