Thursday, 19 July 2018

Annual Report to Stakeholders 2017-18

Summary: 

The last year has been a lot like the year before which works out well mostly. There are a few things I still haven't done that I wanted to which makes it feel a bit like stagnation but let's focus on the victories instead.

Having more professional experience has allowed me to do many of the same things professionally as last year but on a larger scale and faster. For example I taught 6 classes this year, up from 5 last year. I wrote a lot more material, edited a lot more papers, and read a lot more books.

Personal: 

Gabriela and I are getting married in the fall! We get along like a glass... and milk. Stay tuned for next year's report when I will announce the obvious Next Big Step for our budding family, which is obviously getting another dog!

Learning:

Most of my educational reading is still related to scientific writing and technical writing and Publishing. There have been some books on chess, its history, and its variants as well as a lot of skimming of various topics. I've started an annotated bibliography for any books that I've found worth mentioning, but it's going to take some time to populate.

Teaching.:

In the last year I've taught 3 new classes: 440, 410, and 300. As well as three repeats: 201, 300, and 342. Each of these new courses was a qualitatively different challenge. Statistical writing (300, mentioned twice) will be address that in its own section.

Stat 410 was sampling Theory, this was the most core course I have ever taught. That is there was much more theory than in the service courses or the programming or writing courses that I've taught otherwise. I stuck pretty closely to the textbook to avoid screwing up too badly, which usually worked. It went marginally well. It was humbling to realize how much worse I've gotten at proofs since my Putnam days 10 years ago in undergrad. Use it or, uh... fail to retain it I guess.

Stat 440 was "Learning with Big Data". I jumped at the chance to teach this course because it was finally my opportunity to present all those cool techniques like the LASSO and neural networks and parallelization and multiple imputation that are all the rage with the kids these days. It was also a chance to use a lot of the lesson plans from the data preparation course I had made. In retrospect, enthusiasm can only get you so far and this course deserved a specialist. I was out of my depth.

Since then a specialist in big data has been hired in the department to teach it and develop big data skills within the faculty, for which I'm grateful and I'm looking forward to seeing what he brings. On the bright side it was a fun class to teach. While doing some exploratory data analysis we found some literal police reports of dumpster fires. One student review came back and said it was the most valuable course of their entire degree, so that's a major win.

As for research,  there hasn't been much progress in the classical sense. The ArXiv preprint that was submitted last year paper was submitted to a journal. A paper submitted long ago also got published.

Service:

One paper and one book were peer-reviewed.

14 more papers were copy-edited for the Canadian Journal of Statistics up from 7 last year.

Also, I've recently signed on to the editorial board with the replication-focused Journal of Metapsychology as a statistical editor.

Publishing and Stat 300:

Stat 300 is statistical writing. It's a course that I've been spending a great deal of time preparing material for. So much so that the material has now become a small book or a large course pack. As of April this year I have completed a minimum viable product of 55,000 words. And material from the second offering of the course is being added right now, by the time the semester is over a publicly available edition with 75,000 words should be available on Top Hat marketplace around September.

There have been two challenges unique to Stat 300 for me. The first was finding relevant material to teach. This has been mostly solved with extensive note writing and preparation. Somehow, I didn't hear about "The Chicago guide to Writing about Multivariate Analysis" by Jane Miller until it was too late to use for the first semester of teaching. I also cobbled together lessons from established scientific conventions and examples from technical manuals, especially for topics like math-language integration.

The Second Challenge was evaluation. I was so dissatisfied with my evaluation methods that I went back and regraded every single assignment. Compared to the mathematical and procedure-based work I'm used to grading, this class felt extremely subjective. Doing all this took 6 weeks, and there were some unintended consequences of posting such a massive re-grade so late into the summer.  However, now I think I've developed a sense for what constitutes an A, B or C grade level student manuscript. There will be examples in the textbook because I expect other statisticians will encounter similar evaluation challenges.

This year I also posted 24 blog posts, up from 20 last year. Traffic to the blog has grown by about 50%. I also established a twitter presence of about 170 followers for networking and self promotion.

Game design:

I've delved farther into the 'why chess' question as well as played about 1,000 matches of rapid and Blitz chess. The shortest non-trivial answer I have so far is that chess was picked up by the expanding Arab Empire at a time late enough for trade and communication but early enough to get a head start on other strategy games. It was originally more popular to play with dice rules and with movement more like that of Parcheesi and Backgammon.

The ability of Chess to be played with abstract pieces instead of figures of humans or animals also made it viable across the ancient Islamic world.

Dice based variants of Chess and its predecessor Chaturanga may have given it the initial kick, but the pure strategy versions gave it the longevity. Support material such as strategy guides and tactics puzzles contributed greatly to congealing it into a game with unchanging rules. Having support material like tactics puzzles makes changing the rules of a game more difficult because it renders the support material obsolete.

One final shift to today's version of Chess, which was called "Mad Queen's Chess" in its early days was already in progress when it received a critical push to update the game to reflect that a queen, specifically Spain's Queen Isabella, could hold such vast power.

I've also attempted the Jeopardy online test again and failed again but by a much smaller margin this year. I also won the trivia game HQ once.