For analyzing the time complexity of a recursion algorithm, we can use a recursion tree to help. By rough calculation, we can divide the computing work into two parts, those at the leaves (bottom level) and those at branches including the root (uppper levels). Depending on the asymptotical positive function the recursive fork complexity may take, we can decide the total big $O$ of the algorithm from one of the three scenarios.
