Solved – Correlogram in R like in Stata?

You can calculate the threshold significance for a given coefficient. To check this, I just did a quick google search. Found this https://stat.ethz.ch/pipermail/r-help/2009-August/207266.html The code to make the calculation is below. I hope this helps.

dev.new(height=6, width=4.5) par(mfrow=c(2,1), mar=c(3,4,0.5,0.5))  testAR1 <- arima.sim(n=300, list(ar=c(0.5))) test_acf <- acf(testAR1) test_acf_CritVal <- qnorm((1 + 0.95)/2)/sqrt(test_acf$n.used) #change the 0.95 to the desired CI abline(h=test_acf_CritVal, col="red") #add the calculated CI to the plot, just to check that it goes in the right spot  test_pacf <- pacf(testAR1) test_pacf_CritVal <- qnorm((1 + 0.95)/2)/sqrt(test_pacf$n.used) #change the 0.95 to the desired CI abline(h=test_pacf_CritVal, col="red") #add the calculated CI to the plot, just to check that it goes in the right spot 

Now that you have the critical value, you can now compare that to the estimated coefficients for the different lags.

Test_acf_lags <- as.vector(test_acf$lag) Test_acf_coefs <- as.vector(test_acf$acf) Test_acf_Signif <- as.integer(Test_acf_coefs>test_acf_CritVal)  Test_pacf_lags <- as.vector(test_pacf$lag) Test_pacf_coefs <- as.vector(test_pacf$acf) Test_pacf_Signif <- as.integer(Test_pacf_coefs>test_pacf_CritVal)  R_Correlo_Table_acf <- data.frame("Lag"=Test_acf_lags, "acf"=Test_acf_coefs, "Signif_acf"=Test_acf_Signif) R_Correlo_Table_pacf <- data.frame("Lag"=Test_pacf_lags, "pacf"=Test_pacf_coefs, "Signif_pacf"=Test_pacf_Signif) R_Correlo_Table <- merge(R_Correlo_Table_acf, R_Correlo_Table_pacf, all=TRUE)  print(R_Correlo_Table) 

EDIT: Added the calculation and plotting of confidence bands for model ID

dev.new(height=6, width=4.5) par(mfrow=c(2,1), mar=c(3,4,0.5,0.5))  Nsim <- 100 test2AR1 <- arima.sim(n=Nsim, list(ar=c(0.85))) test2_acf <- acf(test2AR1, ylim=c(-1,1)) Nlag <- length(test2_acf$lag) Bart_se_acf_frist <- 1/sqrt(Nsim) Bart_se_acf_rest <- sqrt((1+2*cumsum(test2_acf$acf^2)[-Nlag])/Nsim) Bart_se_acf <- c(Bart_se_acf_frist, Bart_se_acf_rest) ConfBands <- qnorm((1-0.05/2)) * Bart_se_acf lines(test2_acf$lag, ConfBands, col="red") #add the calculated CI to the plot, just to check that it goes in the right spot lines(test2_acf$lag, ConfBands*-1, col="red")   test2_pacf <- pacf(test2AR1, ylim=c(-1,1)) Nlag <- length(test2_pacf$lag) Bart_se_pacf_frist <- 1/sqrt(Nsim) Bart_se_pacf_rest <- sqrt((1+2*cumsum(test2_pacf$acf^2)[-Nlag])/Nsim) Bart_se_pacf <- c(Bart_se_pacf_frist, Bart_se_pacf_rest) ConfBands <- qnorm((1-0.05/2)) * Bart_se_pacf lines(test2_pacf$lag, ConfBands, col="red") #add the calculated CI to the plot, just to check that it goes in the right spot lines(test2_pacf$lag, ConfBands*-1, col="red")