
6# melanda
谢谢你。
呵呵,不过我可能没太说清楚我的意思。
我主要是想知道有没有什么理论可以支持这个说法。
或者如果给出一堆数据,要用线性模型和非线性模型来做,为什么要选线性模型?不考虑模型复杂程度问题,怎么去比较这两种方法的优劣。 [/quot
That's theory for model selection. You should read some books about model selection.
If your models fit data well, then choose the simplest one. If you review the definitions of AIC or BIC, you'll find some clue.
不考虑模型复杂程度问题 is 200% wrong by model selection principle.
Why do you need model ? What's the main tasks of statistics? Ask yourself. If you don't know, then read some related books.
Bless and Cheers


雷达卡
京公网安备 11010802022788号







