"为了减少遗漏变量问题，可以添加大量的控制变量" Statistics analysis is not widely guess. It is bonded theory, business and data.
"但同时会带过过度控制问题，这个问题在统计上问题不大，但会损失样本及有限性。" What does this mean?
"对于过度控制，有没有相应的模型评判标准" there are many ways to test overfitting, from a nested model F test or likelihood value based test to specification test. But the model should be the first built under the guideline of theory and business.
扫码加好友，拉您进群
全部版块
我的主页
收藏
