It is shocking to me that a statistics department would offer a graduate-level statistical computing course only *every fourth year*.

I had been arguing for a statistical programming course: that we supplement the usual course on the *theory* of statistical computing (numerical linear algebra, EM algorithm, MCMC, etc.) with a course on the *practice* of statistical computing.

But I was assuming that the more theoretical statistical computing course was *actually being taught*.

If a department teaches a course at a frequency less than every-other-year, it’s unavailable to many students, or it comes too late in their training to be useful. And statistical computing should really be considered part of a statistics department’s *core curriculum*.

Update: As you might have anticipated, I’ve been asked to teach the course.

19 Mar 2014 at 7:21 am |

I would be very surprised if statistical programing and computation was not an integral part of every PhD statistics course offered.

From my grad school (a different program altogether), I do not remember a lecture and/or the corresponding assignment that did not have a major programing component to complement the theory. If I remember correctly there was only one course on computational statistics, which was rather theoretical and started my love of asymptotic statistics.

Also, when I went to school, it was still common for a master’s level statistician spend most of their second year being trained statistical programing and data wrangling- basically data science these days.

19 Mar 2014 at 7:29 am |

PS Just found your blog while reading Fienberg’s http://www.annualreviews.org/doi/abs/10.1146/annurev-statistics-022513-115703 .

I am really enjoying it. Thank you!

19 Mar 2014 at 7:50 am |

Interesting; thanks for pointing me to that article.