SAS History: I began using SAS® for statistics (SAS/STAT®) back in 1984, during my undergraduate days in zoology at (University of Toronto). I had to. There were analyses I just couldn’t do in other packages. My knowledge of SAS and statistics grew in graduate school and while on a postdoctoral fellowship, where I learned to apply a variety of linear and time series models to experimental and observational data for ecological research. Since arriving at SAS over a decade ago, I’ve filled in many holes and extended my SAS capabilities in data mining, forecasting and even, dare I say, programming.

Partner/Family: Wife – Carolyn, and two daughters – Quinn and Arielle.

Pets: We have two Maine Coon cats, Bear and Willow, and some tropical fish.  

Sports/Hobbies: I mostly play soccer these days, but am a bit of a generalist when it comes to sport – i.e. jack of many, master of none – with volleyball, badminton, cycling and hiking amongst my old favourites. It’s been a long while since any serious hiking excursions – (while living in Vancouver I’d head into the Coast Mountains most weekends), but as my children get older I hope it’s only a matter of time. My top three trips include both British Columbia and Ontario. I’ll never forget that weeklong crawl down the Stein River Valley in British Columbia. The landscapes up beyond the tree-line are amongst the most striking I’ve ever seen, and there’s nothing quite like the walking for days without seeing other people – and the fear one feels while getting lost in the process. Then there was that nasty winter backpack through the Cascade Mountains from BC into Washington State – in the days when our border was thin. It was spectacular but I’ll never do another icy jaunt like that again. Southern Ontario is not BC. But a “stroll” from Rattlesnake Point (Milton) to Collingwood along the Bruce Trail helped me recognize how much we have close by. There are many hidden gems along our Niagara Escarpment if you have the time to look closely. Because of that trip, I’ve returned to the Escarpment many times.

I’m a Torontoist. Is that a hobby? In my lifetime I’ve watched Toronto grow to become a unique world city. I enjoy urban life and think Toronto is one of the great places to be. Toronto history; arts and culture; architecture; neighbourhoods; and urban planning, transportation and development are all of interest to me.

I care about our ravines and urban forest. I volunteer for LEAF and am organizing a tree inventory with City Councillor Joe Mihevc and local residents to help preserve the old growth forest of oaks in our Wells Hill neighbourhood.

I enjoy creative writing. It’s a hobby I picked up later in life, and now can’t put it down.
My eco-fantasy novel Southcrop Forest has won awards for Nature & Environment writing, as well as Fantasy.

Favourite Foods: Indian, Chinese (Dim Sum).

What your ideal weekend would be: I’d like nothing more than to return to Tanzania – this time with our children – and spend a weekend watching the migrations in Serengeti National Park. We’ll camp the first night to hear the hyenas and lions up close, and feel like prey. We’ll spend our days on game drives with field guides and binoculars. A Saturday night stay at a luxury park lodge, with cocktails and savannah view, is a must.

If I could be anything at all (besides a SAS programmer), I would be: Well … I do enjoy variety. Working for SAS has given me the opportunity to meet interesting people from many different industries; work in a variety of roles in training, consulting, pre-sales, customer value, and course R&D; and visit towns and cities across Canada and US. Whatever else I’d do would have to offer as much variety.

Having worked in the sciences for so many years, if I could be anything else (and actually make a living from it) I’d try a career that stretches the other side of my brain – maybe in the visual arts, journalism or creative writing. Of course, if I had chosen such a path and you asked me the same question, my answer might very well be “scientist or statistician.”

When I’m not programming in SAS, I like to: Canoe or kayak; read the morning paper, the Sunday Times and the New Yorker – where the best writers write; watch good and bad action flicks; take the kiddies for Dim Sum; stroll through old downtown neighbourhoods like the Annex or new hipster-hoods like Ossington/Trinity Bellwoods; browse galleries in Yorkville and Queen West; take a Toronto Tree Tour (www.treetours.to/).

One thing every SAS programmer should know: GLMSELECT is a useful but often overlooked procedure. It behaves likes a combination of REG and GLM and also allows for empirical validation and model tuning. The procedure can be downloaded for SAS 9.1 from the SAS Web site. It is included in SAS/STAT 9.2 software. See below for sample code and explanation.

ods graphics on;
proc glmselect data=analysisData testdata=testData
seed=1 plots(stepAxis=number)=ASEPlot;
partition fraction(validate=0.5);
class c1 c2 c3; 
model y =  c1 c2 c3 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10
x11 x12 x13 x14 x15 x16 x17 x18 x19 x20
/ selection=stepwise(choose = validate select = sl);
run;

The ODS statements allow access to statistical graphics requested with the PLOTS option. (Note that ODS graphics are experimental in SAS 9.1 and production in 9.2.) STEPAXIS=NUMBER will plot the model step number (and terms in the model) on the X-axis. The ASEPLOT option will show average squared error for training, validation and test data on the Y-axis. A sample plot is shown below.

The TESTDATA= option allows the user to supply a TEST data set. The SEED= option sets the seed for random assignment of the DATA= data set into training and validation, requested with the PARTITION statement and FRACTION option. VALIDATE=0.5 will partition 50 percent of the data for training and 50 percent for validation.

CHOOSE=VALIDATE on the STEPWISE option will select the model with the smallest validation average squared error, from a sequence of model steps (in a stepwise regression). SELECT=SL will generate these model steps using the traditional approach of statistical significance.