parametric method assumes more, if the assumption is true, more inference can be done. If the assumption (specification) is wrong, then it makes less sense. But, there are quite a few misspcification tests such that one can check for the correctness the assumption made.
with nonparametric approach, one assumes less such that he would never make any mistake. The point is that less inference from the results can be done than with parametric methods.
It may difficult to use two line to give you a 简单的例子. try to read some stuff on kernel density estimation, ..Nadaraya-Waston (NW) estimation. You aslo need to be very good in statistical inference for you to better understand the difference here, at least I think.
加好友,备注jltj
京公网安备 11010802022788号
论坛法律顾问：王进律师
知识产权保护声明
免责及隐私声明
发帖
雷达卡
提升卡
置顶卡
沉默卡
变色卡
抢沙发
千斤顶
显身卡
