フリー問題

問題文

A group compares two algorithms for a table where the effect of one field depends on the value of another field and the boundary is not a straight line. Which choice fits and why?

選択肢

1. A tree based method fits because the fit always needs fewer passes over the rows than any other family of algorithms.
2. A linear method fits because the weights it learns can be read off as the importance of each field for the whole table.
3. A linear method fits because the weights it learns capture the way one field changes the effect of another field without a manual term.
4. A tree based method fits because the splits it makes capture the way one field changes the effect of another field without a manual term.