Fellow Monks

I wanted to share with you an article I found in the Journal of Statistical Software. The article is titled Using Perl for Statistics: Data Processing and Statistical Computing and was written by Giovanni Baiocchi. In this 75- page article, the author provides a nice introduction to Perl and describes the use of Perl for Statistical Computing. There is also a section on The Perl Data Language (PDL). Here is the abstract of the article:

“In this paper we show how Perl, an expressive and extensible high-level programming language, with network and object-oriented programming support, can be used in processing data for statistics and statistical computing. The paper is organized in two parts. In Part I, we introduce the Perl programming language, with particular emphasis on the features that distinguish it from conventional languages. Then, using practical examples, we demonstrate how Perl’s distinguishing features make it particularly well suited to perform labor intensive and sophisticated tasks ranging from the preparation of data to the writing of statistical reports. In Part II we show how Perl can be extended to perform statistical computations using modules and by “embedding” specialized statistical applications. We provide example on how Perl can be used to do simple statistical analyses, perform complex statistical computations involving matrix algebra and numerical optimization, and make statistical computations more easily reproducible. We also investigate the numerical and statistical reliability of various Perl statistical modules. Important computing issues such as ease of use, speed of calculation, and efficient memory usage, are also considered.”

The Journal of Statistical Software is listed in the Directory of Open Access Journals. For information on Open Access, see the Open Access entry in the Wikipedia




Update 1:

Fixed typo (s/Perl Programming Language/Perl Data Language/), as pointed out by Jenda.

  • Comment on Recommended journal article on the use of Perl in statistics

Replies are listed 'Best First'.
Re: Recommended journal article on the use of Perl in statistics
by davidrw (Prior) on Nov 25, 2006 at 16:27 UTC
    In this 75- page article, the author provides a nice introduction to Perl
    ... that never once includes the words strict or warnings :(
    (I noticed it recommends Switch, too; see using switch)
Re: Recommended journal article on the use of Perl in statistics
by brig (Scribe) on Nov 27, 2006 at 21:47 UTC

    Thanks for the post, this is truly useful info to anyone that works with sci code.

    The author appears to work in a Windows™ environment. This seems natural since the statistical word is filled with Windows™ software. I am thrilled that a native of the environment recognizes the power of a CLI and tools and strongly recommends it as useful.

    I agree with another response that there is no comment on -w, use strict, or use warnings (I searched) but again note that this is from a Windows User™ and they don't have shebangs for one... I would agree that repeating these lines over and over in the code would be absurd, but they should at least get mentioned.

    On the other hand the code is actually quite clear and with a casual glance (75 pages 10 minutes ;) appears to closely follow many of the recommendations of "Perl Best Practices". Perhaps a perl monkey with more seasoning than myself can shed more light on this.

    One of the main points of the article is that there are numerous, easy to use extensions to Perl. This is developed by using several of them.

    There are comments on what your next module project should be if you know stat maths, and the recommended testing methodologies with references. (The author simply states what is missing ;)

    I plan to read it tonight and spend more than 10 minutes on the effort. If this thread shows any interest I'll post back anymore details I discover

    The author does in fact make one very bad mistake, and that is to fall for the old trap of thinking that the pretty name for the suits (Practical Ext...) is in fact the real name. Perl is actually Pearl with a lazy spelling.

    Note: When used as an acronym Perl actually stands for: Profoundly Euphemistic Ravisher of Linguistics

    Love, --Brig