Q&A with a Prospective Student (Dec. 2014):

Q:  I noticed your website hasn’t been updated in a few years.

A:  Ah, the website.  Generally I am having a lot of fun with other things, and thus don’t devote a huge amount of time to the website.  On the flip side, students do look at such things, so you have motivated me to update it.

Q:  Are you  still working on data analysis in non-standard spaces?

A:  Yes, I am still working very actively with data in non-standard spaces, and also novel data analysis approaches, such as Topological Data Analysis.  You can see what I’ve been doing in a course this semester, which is just finishing:
Or check out a recent survey paper I’ve written, which is attached. This is now published as:

Marron, J. S., & Alonso, A. M. (2014). Overview of object oriented data analysis. Biometrical Journal.

Q:   Your research interests mention image analysis- What particular  types of images do you typically work with? I have experience  working with fMRI and 3D images and was wondering if my experiences overlap with your work.

A:  As a statistician, I don’t focus a lot on particular image flavors, but tend to collaborate a lot with imaging experts, who handle the image detail work.  Instead I more enter the game at the population level, and usually find I need to invent new statistical methods to take on the challenge appearing at that level.  But to answer your question, lots of different imaging modalities appear in papers I have published, from H&E microscopy to MRA (for Angiography).  Lots of these are 3d, from various MR modes (T1,T2, MRA, DTI,…) to various CT modes (straight, trace enhanced, cone beam…), and others (PET, Gamma Camera…).  I have not personally done a lot with FMRI, but our department has others who actively work in that area.

Q:   The application of your work spans everywhere from biology to marketing. Are the same methods applied to all types of data or do the methods need to be altered to fit the specific application you’re addressing?

Q:   This is an interesting question, that does not admit a simple answer.  I do work in a mode of trying to find specific answers for people addressing real problems.  But as a statistician with close collaborators in a wide range of areas, I also view my work as being about cross-fertilization, and taking great ideas from one context into another one.  One way I try to do this is to find common general ideas in all these contexts.  E.g. this what the OODA idea in the referenced paper is all about.  Now how close is this to “using the same methods in different contexts”?  Perhaps that happens surprisingly infrequently. Almost every time I tackle a complex problem, new statistical ideas are needed.  Of course they draw a fair bit from previous ideas and experiences, but always seem to need new stuff woven in as well.  Many of the PhD dissertation I currently supervise come form taking a domain problem, working with the data to understand what the needs are, then coming with an innovative solution.  A really nice current example of this mode of operation is in virology, and the invention of Radial DWD.

Q:  Do you personally enjoy working in academia?

A:  I love my academic job.  I have a lot of fun every day, and they pay me to do it.  My current wife (lost my first to cancer a few years ago) enjoys remarking on how often I tend to enjoy the events from each day (not always great of course, but that is life).

Q:  Do you believe that UNC’s program prepares students for both industry and academia or is there a stronger pull to one direction?

A:  My students have a great record of finding jobs to their liking.  They can be found in all of (they can generally choose, and I try to tailor what they do in the dissertation somewhat in the direction of their ultimate goal) academics, industry and government.