Georg M. Goerg
I am a PhD student (ABD) at the
Department of Statistics at
Carnegie Mellon University, where I
am working towards presented my thesis proposal (Spring 2011) on nonparametric prediction, complexity measures, and automated pattern-recognition of space-time processes (random fields), and the statistical theory thereof. My advisors are
Cosma Shalizi and
Larry Wasserman.
(Reproducible) research interests: (long memory) time series, skewed and heavy-tailed distributions, Gaussianizing and exponentialize data, time series forecasting, spectral analysis (of time series/signals), locally stationary time series; Lambert W function and Lambert W distributions.
CV GMG; and here is a more detailed account of mea vita.
Due to great feedback on my JS 2011 presentation on the nonparametric frequency domain EM algorithm I put the pdf online: Frequency Domain EM Algorithm - JSM 2011. And here is the link to the manuscript - which also got published at SAM.
Winner of the JSM 2012 student paper competition in Statistical Learning and Data Mining: Locally-stationary Box-Ljung test (new, reviewed version for !)
New R package ForeCA Sunday, 20 May 2012
Along with my new paper on Forecastable Component Analysis (ForeCA) I also put a new R package on CRAN: ForeCA. It currently has basic functionality to estimate simple quantities such as spectral entropy or from time series. It also provides nice wrapper functions to work with multivariate spectra of multivariate processes.
All good things come in threes: 3rd time student paper competition winner now in JSM 2012 Thursday, 26 January 2012
Driven by my competitive side I digged up a manuscript hidden for a long time on my hard disk entitled Testing for white noise against locally stationary alternatives. After some days to polish it up, I submitted it to the 2012 JSM student paper competition held by the Section of Statistical Learning and Data Mining, sponsored by the journal with the same name (SAM). And to my - positive - surprise my paper was selected as 1 of 5 winners - just like last year, and in 2007. San Diego here I come. Update: pdf onlne . To be published in SAM
Nonparametric Frequency Domain EM for Time Series Classification published at SAM Monday, 03 October 2011
My student paper competition entry on time series classification using a frequency domain version of the well-known EM algorithm was not only awarded as one of the 5 winners at JSM 2011, but was also published in Statistical Analysis and Data Mining - SAM. The manuscript can be found at arxiv.org/abs/1103.3300.
