Hello, I'm a novice R user. I'd like to estimate the linear trends (b) and their statistical significance (p-value) of quite a few univariate time series. Because several time series show autocorrelation and heteroskedasticity, the ordinary least squares method lm(), as I understand it, isn't the most appropriate choice; the generalized least squares method gls() in package nlme can account for both autocorrelation and heteroskedasticity, and therefore seems to be my best shot. But some of my time series also contain outlier observations, which gls() might not be robust against.
I'm now playing around with the robust linear regression estimator lmrob() in package robustbase. According to http://cran.r-project.org/web/views/Robust.html, lmrob() "uses the latest of the fast-S algorithms and heteroscedasticity and autocorrelation corrected (HAC) standard errors". However, such information is absent in the lmrob() help. I wonder if anyone is able to confirm or repudiate the application of lmrob() to autocorrelated and/or heteroskedastic time series. Again, my goal is simply linear trend estimation. Thanks, David [[alternative HTML version deleted]] ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
