how to get residuals and perform the Durbin Watson test on Rolling Linear regression?-CodePudding

i have run this below coded for which I got the result of coefficients and r2 values,

model <- roll_regres(y ~ x1   x2   x3   x4, Data, width = 15L, do_downdates = FALSE, do_compute = c("sigmas", "r.squareds"))

coef <- outneg$coefs

r2 <- outneg$r.squareds

and what I need is to do the Durban-Watson test on residuals? the problem here is I am not able to extract the residuals from roll_regres. how do I get the residual values for rolling window regression?

CodePudding user response：

Do you mean the residuals of each of the expanding length window regression model? You can get them like this.

residuals = lapply(15:nrow(Data), function(x) {
  lm(y ~ x1   x2   x3   x4, data = Data[1:x,])$residuals
})