A midway review of my professional life in 2018statistics graduate school productivity
“Hindsight is 20-20,” at least that’s how the saying goes…
As of a few weeks ago (Friday, 11 May 2018), I find myself with yet another year of graduate school behind me — and perhaps now is a good time to ponder how the last year has changed me. Surely, I’ve gained in my statistics education (maybe?) but what of my goals? Do I still want the same things as I did just last year? For now, it seems that things have changed considerably.
At the beginning of this past academic year, which marked my first in the PhD program in Biostatistics (and my second year of graduate school), I wanted — badly wanted — a future career in the academy, alongside expertise in methods at the cutting-edge of computational statistics. Now, I find myself dissatisfied with academia — in large part due to the sort of individual conduct that is actively tolerated and the personality politics that seem to dominate the landscape — and find my interests and goals better aligned with what seems to be more highly valued in industry research settings. Really, what’s the point of new and exciting theory if it never sees a chance to shape the world? For this, we need to actively encourage and reward the development of robust software tools (at least in Statistics). Unfortunately, such an attitude has not yet been realized, though the data science push has moved us closer. Where before I found myself extremely interested in computational methods, I now find myself drawn once again to the applications of statistics in science (having decided to pursue a graduate minor in “computational and genomic biology” in place of one more focused on “data science”). Though I’ve yet to give it another try, it may well be that the same issues of personality politics exist in industry settings as well — who knows?
For all these apparent changes in attitude and goals, much has remained the same — I still find myself interested in issues at the intersection of causal inference and machine learning, nonparametric statistics, and the myriad applications of these tools in the sciences (with a rekindled interest in computational biology due to a seminar with Nir Yosef) and policy (due to a chance to discover issues in algorithmic fairness through a seminar with Moritz Hardt). Over the last year, I’ve started working much more closely with David Benkeser (now at Emory) on a project broadly concerned with causal inference using stochastic interventions and Mark van der Laan (now my primary advisor) on topics ranging from the highly adaptive lasso to differential methylation analysis; moreover, I’ve also continued working with Nick Jewell (soon to move to the London School of Hygiene) on a project concerning survival analysis in settings in which careless analyses may be plagued by immortal time bias while a methods project (on applying Bayes shrinkage to Targeted Learning) and applied work with Alan Hubbard has largely been left to stagnate. A current hope is that several of these methods projects will be ready for publication by the end of the present summer.
These ramblings seem to be getting a bit too long-winded, so now seems a good place to stop. I’m hoping to follow up with a post to set goals for my next year in graduate school soon, before the summer’s end. In the mean time, I plan to dedicate time to posting reading notes on causal inference and topics related to stochastic interventions on my statistics blog as preparation for my qualifying exam, to take place on Friday, 14 September 2018.