Statistical computing for astronomy with R

Eric Feigelson (Penn State University)

Date: Sunday 6 November 2011, 14:30 - 17:30 (exact time TBC)

Abstract

R (http://r-project.org) is the world's leading pubic domain statistical analysis software environment. A high-level language similar to IDL with a C-like syntax, R provides a flexible data analysis capabilities and diverse high-quality graphics. Most importantly, R allows ready implementation of hundreds of functionalities from all branches of applied statistics including statistical distributions, density estimation, nonparametrics, least squares and maximum likelihood estimation, hypothesis tests, multivariate analysis, time series, and survival analysis.

R is designed for interactive data analysis, but can be integrated with previous codes. R can ingest external subroutines in C, C++, Fortran, Python, Java, Perl, Ruby, and other languages. In the other direction, R functions can be called from C, Fortran, Python and Perl. We are now designing a Web service interface called VOStat providing a limited suite of R functionalities within the Virtual Observatory system, and plan other software infrastructure codes for astronomy.

R is open-source and has invited users to provide add-on packages implementing specialized statistical packages through the Comprehensive R Analysis Network (CRAN). CRAN packages are introduced into an R session on-the-fly from Web mirror sites. CRAN has expanded exponentially since 2001; in 2011, over 3000 packages are now available and a new package arrives every day. These provide a vast range of advanced statistical functionalities: Bayesian inference, machine learning and data mining, wavelet analysis, spatial point processes, image processing, high-performance computing, and much more.

Note for participants

Participants should bring their laptops for interactive use of R during the tutorial.  R can be installed quickly from http://r-project.org, and R scripts for the tutorial will be provided on a Web site for interactive operation

Author Biographies

Eric Feigelson is Prof. of Astronomy & Astrophysics at Penn State and Associate Director of Penn State's Center for Astrostatistics. He has co-organized summer schools and conferences in statistical methodology for astronomy, and is lead author of the forthcoming volume `Modern Statistical Methods for Astronomy with R Applications' (Cambridge Univ Press).