r - Stata 与预测的标准差和 R 与预测的标准差有什么区别？

Question

据我所知，Stata和 R具有“预测”功能。我正在尝试使用 R 复制在 Stata 中执行的结果，结果涉及计算预测值的标准偏差。 R 中是否有一个功能，也许使用它的“预测”功能，可以让我这样做？我似乎无法完美地复制结果。如果有帮助，Stata 代码会执行以下操作:

reg Y X1 X2 if condition
predict resid, r
predict stdf, stdf

stdf 参数的定义是:

stdf calculates the standard error of the forecast, which is the standard error of the point prediction for 1 observation. It is commonly referred to as the standard error of the future or forecast value. By construction, the standard errors produced by stdf are always larger than those produced by stdp; see Methods and formulas in [R] predict

我一直在写的 R 代码是:

fit <- lm(Y ~ X1 + X2, data=df)
new.df <- data.frame(...) # This is a new data frame with my new data period I want to predict in
predict(fit, new.df, se.fit = TRUE)

但是，当我将标准误差转换为标准偏差时，它们与 Stata 输出不匹配。

提前致谢!

Answer 1

在我看来你需要:

 predict(fit, new.df, se.fit = TRUE, interval="prediction")

“标准误差”适用于均值估计值的置信限度，而预测误差可能很容易描述为预测值的“标准偏差”。

> dfrm <- data.frame(a=rnorm(30), drop=FALSE)
> dfrm$y <- 4+dfrm$a*5+0.5*rnorm(30)
> plot( dfrm$a, predict(mod) )
> plot( dfrm$a, predict(mod, newdata=dfrm) )
> points( rep(seq(-2,2,by=0.1),2),   # need two copies for upper and lower
          c(predict(mod, newdata=list(a=seq(-2,2,by=0.1)), 
                         interval="prediction")[, c("lwr","upr")]), 
          col="red")
> points(dfrm$a, dfrm$y, col="blue" )