Thursday, May 10, 2012

Record-Setting Attendance for A Record-Appreciating Audience

Of course, I’m not referring to LPs and EPs here… that would be something this former Disc Jockey would certainly gravitate towards but I’m quite sure more of this blogs audience is actually more into data records.   Actually, come to think of it, so am I in the professional phase of my life!

I’m sitting in an airplane as I write this.  I know: what a HUGE surprise.  To tell you the truth, I do feel a little discombobulated at the moment.  Wednesday morning I was up bright and early to head in to the office in Toronto – with a little trepidation, I might add – for the Toronto Data Mining Forum and by dinner I was in Saskatoon.  I was feeling a little anxious a little while ago for all the right reasons.  Registration for the Toronto Data Mining event was hovering around 350 individuals, and we only had enough room for about 260.  Even with expected drop-off, it was a little nail-biting for me as I never want anyone to be turned away from a SAS Canada user group… that would certainly have been a first.

Another important first was actually achieved, however.  I’ll have to confirm this for sure, but by scanning the room, I could guess-timate that there were about 220 people in attendance.  I know, I know, it’s exceptionally dangerous to use the work ‘guess’ around statisticians and data miners: it’s practically a swear word in some circles.  This attendance for a single, half-day morning session (if accurate) will represent the single largest gathering of SAS users for a user group in Canada ever.  All credit goes to the executive committee led by Dina Duhon of Scotiabank who so consistently find incredible presenters offering great content to a willing and eager audience.  Congratulations to them for helping to realize this amazing achievement! 

The agenda certainly promised great things.  The presenters were all relatively unknown on the stage of the user groups, but their topics were certainly compelling.  Given the audience reaction to each of the talks, it was clear to me that they were a hit.  This just goes to show that anyone can – and should – feel empowered to stand up and give a talk on any topic at a user group meeting.  Nothing but good things can happen.

Leading off the agenda, Charles Chen started us off with an hour on
survival model attrition and analysis.  With great energy, humour and above all, confidence, Charles walked us through a variety of scenarios and the rationale behind employing survival models within a finance-based environment.  The patient was in fact the customer, and permanent events – possibly death, in a health context – were representative of churn and attrition.  I have to say, the audience LOVED this talk.  Charles was fielding questions left, right and centre with a smile on his face and a well-thought out answer on his lips.  As he finished I had at least five people come up to me raving about his talk and hoping to get an advance copy of it to take back to their environment and start discussions around implementing similar strategies.  There is no greater reward within the SAS community than helping someone else: whether an individual or an organization.  In this respect, Charles more than succeeded.  The best news of all is that he had to leave out a significant portion of information… I guess this means we’ll happily have him back for round two in the near future.  Great job Charles, I know we all learned a lot!

Following our networking break, Masoud Charkabi of CIBC took the microphone.  Masoud is a young, well-polished, unflappable presenter.  He clearly radiated confidence in his material and was no stranger to the stage.  Some people might be intimidated talking to a room full of data mining practitioners about considerations around computing resources in data mining, but not Masoud. 
From both a hardware perspective and a software perspective, he confidently walked us through some of the more important considerations when considering efficiency and scalability.  He even referenced both my own brief talk around SAS’ High Performance Analytics offerings and a previous presentation given by Daymond Ling on segmentation at the fall Data Mining Forum.  I was very impressed with how he tied together so many disparate elements into one cohesive talk…. and he teased us enough about text mining for me to make a note that he would be an ideal person to give a talk around this fascinating topic in the future.  Fantastic stuff, Masoud!

Our next presenter kept up the hot streak of the group in terms of exceptional delivery and relevance.  Mario Wen of TD Bank discussed the
philosophy and application of fraud detection analysis.  Coming from the perspective of debit card fraud, he was able to explain in layman’s terms how fraud takes place and his organizations strategy for fighting back.  He likened it to a game of cat and mouse, and it certainly seemed that way; the ‘bad guys’ were always learning and adapting, but so were the good guys.  Fraud detection is clearly an art form which requires patience, diligence and near-constant vigilance.  Having been on the opposite end – my debit and credit cards have both been shut down erroneously by banks because it was perceived as unusual behavior – I can say that I didn’t mind one bit the slightly over-zealous security measures taken by anti-fraudsters.  I’d rather be protected constantly and inconvenienced rarely than the other way around.  Mario handled all the questions that came his way with unflappable calm, and I was truly impressed with the complexity of his subject – but the simplicity of how he conveyed it.

Finally, Bob Saaramaki of DataMaApp closed out the session with a thought-provoking presentation on
using experimental design techniques to maximize marketing efforts.  Bob is a natural; a great presenter very comfortable and at ease in front of an audience, he was able to explain and contextualize his organizations unique approach to marketing processes with great clarity and humour.  I shouldn’t have been surprised: when Bob presented at the TASS meeting in March, his room was overflowing!  At any rate, I think that he opened some eyes and got some wheels spinning around a much more efficient way to do marketing.  In terms of time, lift, and spend, he very effectively demonstrated that a biology-inspired approach to experiment design could indeed pay off in spades… and could be driven through base SAS as well!  I only wish we had more time for him to dive deeper into his talk.  Next time, Bob.

At meetings’ end, I heard from many attendees that they really appreciated all of the presentations and the general sense of community which came from the group.  Here’s hoping the next meeting is just as well-received!


Until then…

Sunday, May 6, 2012

Business Analytics & The Next Wave of SAS Solutions

I had the privilege this past week of attending the Forum analytique SAS de Quebec and the Forum analytique d'affaires SAS de Montreal, two of the leading edge user groups in Canada.  These groups are formidable in structure and in scope, and last week's meetings were no different.

Both groups are strongly supported by SAS Canada solution specialists which means that they have the opportunity to have the newest SAS solutions presented and demonstrated by some of the brightest in Canada.  The theme for both meetings this past week was high performance analytics (HPA) with a strong focus on the new SAS Visual Analytics solution, which featured heavily at SAS Global Forum in Orlando.  I, for one, was extremely excited to see the talks on the slate.

I've been fascinated by HPA long before Dr. Jim Goodnight demonstrated the unbelievable hardware power required to perform billions of calculations in minutes at SGF... with the aid of a hologram, no less!  I'm one of those data geeks who really loves efficiencies of all kinds.  If a few hours, minutes or seconds can be shaved off a task, I'm immediately interested - although given my proclivity for coffee breaks during this time, perhaps my local barista doesn't necessarily share my excitement given the sharp drop in business which will inevitably follow. 

Both groups featured Ron Allard giving an overview of the challenges of Big Data which make HPA such a game-changer. His talk was entitled 'Le "Big Data": Pourquoi les entreprises devraient s'en préoccuper et quels sont les moyens d'y faire face'? or in English, why organizations need to pay attention and how to do so most effectively.  Ron's talk certainly sparked some great conversation in Quebec where a small but engaged audience found a lot of value out of his talk.  Ron nicely framed out the consideration around what Big Data actually is - no small task in a rapidly changing discussion space.  He also gave an overview of some of the hardware requirements necessary to use HPA effectively and discussed the difference between in-database and in-memory analytics. His talk also nicely set up Andre Lafreniere who would be carrying the majority of the meeting with an overview and demonstration of the SAS Visual Analytics solution.

Now, I've seen VA demonstrated a few times.  I first saw a sneak peek of it at our annual kick-off meeting in January.  Of course, we saw a lot of it at SAS Global Forum as well.  Now I don't know about you, o gentle reader, but I often wonder if the products work as quickly as they do when they're not the feature component of a huge conference or presentation.  After last weeks meetings, I can happily attest that the answer is 'yes'.  I was blown away with how quickly the drag and drop functionality worked.  Andre was able to run logistic regressions against a fairly robust dataset in mere seconds... and the GUI interface was as slick off of his iPad and laptop as it was on the big stages of the past few months.  True, I'm a SAS employee, but I con honestly say I was blown away by the possibilities allowed by the combination of HPA and VA.  No more subsetting data for the sake of saving time?  Visual data exploration, real-time drilldowns and easy drag-and-drop?  Sign me up!

The Canadian SAS user groups are traditionally a great way of supporting SAS users by offering tricks, tips and hints around optimizing existing SAS implementations.  This often means speaking to the vast number of SAS programmers out there who are doing amazing work with a tried, tested and true system.  I'm glad I had the chance to get a glimpse at the future of SAS and analytics... who knows, in 5 years or so, perhaps we'll be seeing tips and tricks on SAS Visual Analytics.  I think the users who attended were impressed by the next wave of SAS solutions and I hope we'll see even more talks as they become prevalent in industry moving forward.

Speaking of moving forward, I have a hectic week of user group meetings coming up.  Toronto Data Mining, Saskatoon and Winnipeg are all on the docket for me this week with my manager Wally Thiessen making the short hop down to Hamilton for the GHSUG user group meeting on Friday.  You'll be hearing a lot more from me over the next seven days or so, I can guarantee you that!

Until then...