R2的值太低怎么办

if R2 is too low, it means that other improtant factor u haven't included in ur regression model. therefore u have to add other factor in order to improve R2 value. However, R2 is a trick, u have to realized that more variables u add in ur regression model the higher R2 u will. as above people said, u can concerned on Singnificant level. Inadditional, if R2  between .5 to .8 u then u don't have to worry abt that

bjmayi 发表于 2009-10-26 14:07
求助, 已经采取措施消除了多重共线性, 做出模型后, 系数变化基本符合经济学规律, 但R2的值太低,  只有0.1左右,  这样的方程还能用吗?

Cross-section data usually has very R2 value. Beacuse a large amount of individual heterogeneity cannot be explained by model.

可能有以下2个原因：1.选择的模型形式不对；2.如果各解释变量的t很大，那么就是模型形式正确，但是选择变量不起解释作用，也就是说选择的解释变量不当。

低的R2，多半是拟合有度过差导致的。
在横截面数据中，通常R2很低，除非问题特别好，线性性吻合比较好。

jingjiren2007 发表于 2009-10-26 18:02
我觉得不能用了，拟合优度太差了，说明方程没有选好，可以另选一个更符合的方程试试吧。

看具体方向，公司金融方向的文章你去看下top5的期刊，0.2就已经非常非常好的了，一般都在0.1左右，有的甚至更低。