The problem with data analysis is of course that it is a performing art. It is not something you easily write a paper on; rather, it is something you do. And so it is difficult to publish.

quoted from this conversation ——————————————————-

Statistical Science has a nice “conversations” series with renown statisticians. This series always benefits me because of 1. learning the history of statistics through a personal life, 2. confronting various aspects in statistics as many statisticians as were interviewed, and 3. acquiring an introductory education in the statistics that those interviewees have perfected over many years in a plain language. One post in the slog from this series was a conversation with Leo Breiman about the two cultures in statistical modeling. Because of Prof. Huber’s diverse experiences and many contributions in various fields, this conversation may entertain astronomers and computer scientists as well as statisticians.

The dialog is available through arxiv.org: [stat.ME:0808.0777] written by Andreas Buja, Hans R. Künsch.

He became famous due to his early year paper in robust statistics titled, Robust Estimation of a Location Parameter but I see him as a pioneer in data mining, laying a corner stone for massive/multivariate data analysis when computers were not as much capable as today’s. His book, Robust Statistics (Amazon link) and the paper Projection Pursuit in Annals of Statistics (Vol. 13, No. 2, pp. 435-475, yr. 1985) are popular among many well known publications.

He has publications in geoscience and Babylonian astronomy. This conversation includes names like Steven Weinberg, the novel laureate (The First Three Minutes is a well known general science book) and late Carl Sagan (famous for books/a movie like Cosmos and Contact) showing his extent scholarly interests and genius beyond statistics. At the beginning, I felt like learning the history of computation and data analysis apart from statistics.

The NR website: http://www.nr.com/

Some noticeable additions are

1. chi-square with small numbers of counts,
2. various statistical applications including Markov chain Monte Carlo, statistical learning (classification and clustering– KD-tree, model based clustering, nonparametric multivariate data analysis), computational geometry, sparse matrices, kriging, and graphs,
3. improved functions including incomplete gamma and beta functions,
4. wavelet on the interval, and
5. information-theoretic properties of distributions.

I’m sure you’ll find more interesting topics to your liking that are added in this new edition. Maximum Entropy Image Restoration, I didn’t know this topic was in Numerical Recipes in other editions. The third edition looks very modern to me compared to the second edition. You’ll enjoy it.