Thursday, October 18, 2012

Interacting on the Catwalk – Modeling Takes Centre Stage at the Toronto Data Mining Forum

This morning the Toronto Data Mining Forum convened at the SAS Canada offices for our semi-annual meeting.  I’ve come to really look forward to these get-togethers for a few reasons.  First, the agendas have become more and more compelling over the past few years and this has been reflected by steady growth in attendees.  In fact, one of today’s presenters had delivered a talk about 5 years ago in the early days of the group to a room of about 20 people…  a far cry from the roughly 130 individuals who packed the room today!  I’m also very happy that the Forum allows me an opportunity to get together with a very strong, very dedicated and very forward-looking Executive Committee.  Without taking anything away from other user groups, the Data Mining Forum consistently addresses the pressing issues in industry today. Today’s meeting featured a strong focus on modeling in several forms.

The topic of modeling can be either extremely complex.  There are many considerations when considering how variables interact: what is important, what isn’t, what has a marked effect on something else, what is skewing results… and these are just some of the more obvious challenges!  Given the broad scope of this topic I was glad to see that today’s meeting really covered most of the bases.

Our first presenter was Ryan Zhao. Ryan is a really interesting person.  He described to me how he had been originally inspired and connected through the auspices of both the Data Mining Forum and the SORA seminar series in the early days of his career. This in no small way factored in his decision to give something back to the data mining community… not only through
his presentation on realizing ROI through uplift in marketing campaigns which he delivered today, but by providing training for new analysts through his business. As Ryan explained the ‘analytical gap’ he was hoping to fill, I couldn’t help but admire his vision and passion for data mining and analytics in general. This passion was certainly reflected in his talk which was very well-received by all.  As a marketer myself, it was very gratifying to see a talk which lend some credibility to the complexity of the task before us on a daily basis… so thanks for making me feel better about myself, Ryan!  I suppose I should pay you for the unintentional therapy session.

Following Ryan, Tim Gravelle of PriceMetrix took the stage. Tim is no stranger to the user group stage having presented previously at the Toronto Area SAS Society and preparing for talks at both the North East SAS User Group (NESUG), a larger regional conference based in the United States… and potentially at the upcoming SAS Global Forum in San Francisco.  His comfort certainly came through in his talk!  Tim discussed the
importance of interactions and how they can help explain behavior in a significant way… but only if they’re given the proper weight and if they’re treated with the respect they deserve.  Using a a traditional SAS programming method, Tim illustrated how deep this subject can get; his knowledge and wealth of experience – especially when he brought to bear his former experiences at Gallup – were well-expressed.  Over lunch, we pressed Tim a bit on his view on the upcoming election south of the border… I won’t reveal his opinions or perspective – that would be cheating – but I can honestly say I have a better view of the upcoming decision for our cousins to the south. 

Finally, SAS Canada’s own Lorne Rothman took the stage to discuss
survival analysis using Enterprise Miner.  He admirably surged ahead in his talk though time was his enemy.  A lengthy break and many questions had cut into his time.  Lorne had no issue ensuring that the large audience received full value for staying the whole length of the meeting.  His talk was comprehensive and direct, referencing the previous presentations as well as showcasing his own expertise with data mining principles and Enterprise Miner.  It’s always a pleasure to have Lorne present – especially in front of a larger group – as there are invariably many of his former students’ in the room. I’m glad he was able to talk at length about a subject area he knows well and for which he feels great passion. Thanks for yet another great talk, Lorne!

I mentioned earlier that the data mining group is very forward-looking.  I should also mention that they are a discriminating, open audience who are always wiling to voice their praise – or damnation – for any given agenda.  This meeting was no different.  A few takeaways for me: definitely work with our IT department to investigate sound dropping towards the back of our large meeting space. It’s no fun for people sitting in the comfy chairs at the back to strain and stress to hear the presentations!  On a similar note, I’m going to investigate sourcing out a hand-held microphone for questions.  Rather than having the presenter repeat the question, this would make things exceedingly clear for all.  I’m very pleased that the overall reviews of the meeting were very favourable.  This can be a bit of an intimidating community; the level of expertise and the massive amount of knowledge in the room can be daunting to say the least. Having said that, there are few more tightly-knit SAS communities to be found anywhere across the country. It’s a privilege to contribute to their professional growth in any way!  If you’d like to see the talks which were given today, feel free to have a look at the
Presentation Archive on the Toronto Data Mining page; you’ll find the latest talks posted at the bottom.

Next for me: I’m off to Saskatoon and Winnipeg for their user group meetings. It’s a whirlwind of a week to be sure… especially with transitioning 3 different time zones in a matter of 2 days. As long as I remember what city I’m in – no guarantees – I’ll be back soon to fill you in on my experiences.

Until then…

No comments:

Post a Comment