Hi Martin, I helped to develop Amelia, so I can try to take a shot. In a non-mathematical way, Amelia works by filling in missing values with imputed values that are consistent with the observed relationships in the data, plus some random noise. Thus, Amelia creates multiple imputed datasets that have no missingness (the original observed cells remain the same across each imputation, but the filled-in values vary from imputed dataset to dataset) and have the same relationships between and within variables as the original observed data. The difficult part of the problem is estimating the relationships of the observed data since it has all of that missing data in it (the dataset looks like Swiss cheese). We use a EM algorithm to estimate these relationships, but those details are (somewhat) less important.
You can find more resources, including a number of papers describing the methods at our webpage: http://gking.harvard.edu/amelia Hope that helps! Cheers, matt. ~~~~~~~~~~~ Matthew Blackwell Assistant Professor of Political Science University of Rochester url: http://www.mattblackwell.org On Mon, Jan 7, 2013 at 6:27 PM, zGreenfelder <zgreenfel...@gmail.com> wrote: > On Mon, Jan 7, 2013 at 4:29 PM, Martin <lh...@gmx.net> wrote: > > Dear all. > > > > First of all, my english isn't verry good, but I hope I can convey my > concern. > > I've a general question about the Amelia algorithm. I'm no mathematician > or > > statistician, but I had to use R and impute and analyse some data, and > Amelia > > showed results that fitted my expectations. I'll have to defend my > choice soon, > > but I haven't totally grasped what Amelia does. I'm particularly > interested in a > > simple as possible explanation in how Amelia imputation works. I've read > that it > > uses a bootstrapping-based algorithm, but how does it chose the values? > > The data had mainly value >0 (chemical concentrations, water temperature > and > > pH-value). > > > > Regards Martin > > I'm pretty new here, but a quick google search suggests that perhaps > http://gking.harvard.edu/amelia (and maybe google translate) > might have some decent pointers for you. > > I poked at the documentation from that site (a pdf file), and it's > quite intense on > the mathematics, you may get more from it than I could. there's also > a link there > to a separate, for thisspecific package/algorithm (it seems to be > called Amel ia II, > I'm assuming this is the same that you used, if I'm off .. sorry about > that) > > HTH > -- > Even the Magic 8 ball has an opinion on email clients: Outlook not so good. > > ______________________________________________ > 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. > [[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.
