I’m speaking at SQL Saturday #120 this weekend in Orange County, CA and it’s the 3rd straight year for me at this event. It’s also the first SQL Saturday I presented at two years ago, so I have a soft spot in my heart for this one. Andrew Karcher, Marlon Ribunal, Thomas Mueller and the rest of the volunteers do a great job putting this one together, and this year they have an all-star lineup of speakers. They have 8 tracks scheduled with 6 sessions per track so there’s going to be a lot of content delivered in a short time frame, with a number of SQL Server 2012 sessions so you can learn more about the new release. There’s also a Women in Technology panel at lunch that I highly recommend, especially to the men. A small fringe benefit of attending is that the conference is held in a junior college with classrooms around a courtyard, registration is outside, and going from session to session means a breath of fresh air in the coastal weather of Huntington Beach. I’m presenting two sessions, the first at 11am on Tabular Models, a hot new SQL Server 2012 topic. The second is right after lunch at 1:30 on real-time data warehousing. Here’s the title and abstract of my sessions. I look forward to seeing you there.
Preparing Data for Analysis as Tabular Models
The new Tabular Model in SQL Server 2012 is being touted as a powerful analytic engine that is faster and easier to implement than traditional cubes using Analysis Service. In this session Carlos Bossy will demonstrate how to develop analytic models using your existing data sources and will discuss best practices for preparing data so that it can be effectively used as a Tabular database. In addition to creating the model, Carlos will show you how to quickly add measures, hierarchies, and calculated columns to provide a rich user experience.
Real-time Data Warehouse and Reporting
This session will present a Database and ETL Architecture that allows you to smoothly move data from your application databases across the enterprise in real-time to your data warehouse. Carlos Bossy will show you how to use SQL Server features such as Replication, Change Data Capture, SSIS and other techniques to extract, transform and load data fast and dependably and view a real demonstration of the system in action. Finally, Carlos will show how to use Reporting Services to present data with a minimum amount of latency.
According to the rules of the game established by Jen McCown for Un-SQL Friday, this post was due Monday June 6. Even though the time has come and gone, I still wanted to write because of the topic – Speaker Lessons Learned. Coming off my last presentation at SQL Rally last month, which I wanted to be perfect but wasn’t, I felt that contributing to the topic would help me put my thoughts together and improve my next presentation, and could help others as well.
My presentation titled The Data Mining Lifecycle wasn’t terrible and the evals weren’t that bad either. I got dinged mostly on time management which I knew I didn’t do well. I had too much material and rushed at the end, skipped a few key points, took too long to start the demos, and answered more questions than usual. I had given essentially the same presentation at the Orange County SQL Saturday in April and I felt I did it as well as I can do it. I made all the key points, had good audience interaction, and finished right on time. But we had 70 minutes, so I knew I needed to cut 10 minutes for SQL Rally and planned for it. I was angry at myself for not handling the shorter time frame since I knew I needed to account for it.
These are the lessons I learned at SQL Rally and at a couple of other events in the first half of this year:
Time Management It’s always difficult thing to make the session finish right on time. Even though I cut 10 minutes from my presentation, I was still short on time because I got 10 more minutes of questions than I did in Orange County. Also, in Orange County I allocated 30 minutes for slides and overview and 40 minutes for demos. When I cut it by 10 minutes for SQL Rally I left 30 minutes for slides and 30 minutes for demos. With the extra questions it turned out to be more like 40 minutes of slides and 20 minutes of demos, which made it more Powerpoint and less of the interesting stuff.
Lesson #1: Make your key points and show demos early so you don’t feel like you need to squeeze them in at the end.
Question Management I got a couple of questions that came out of left field, and one that I didn’t understand at all. I think I let these questions throw me off and I lost my momentum for a minute.
Lesson #2: Be quick to recognize a question that is off-topic or can’t be answered to maintain momentum.
Improvise At a .NET conference I decided I would talk about BI to .NET developers. While I prepared it as a beginner session, I quickly became aware that it needed to be much more beginner than I had planned based on the questions I was getting.
Lesson #3: Be prepared to adjust the presentation for the Audience if necessary.
For the Audience
I have one recommendation for people who are session participants. There were several sessions I attended at SQL Rally where the speaker was struggling. They had done their best to answer a lot of questions, had been challenged by a couple of ‘experts’, allowed for unrelated comments from the audience, and needed to move forward. When you see that happening, help the speaker out. Hold a question for later, or catch up with the speaker at lunch and ask the question then. If you need to discuss your own particular work environment and how it applies to the speaker’s presentation, don’t take up everyone’s time.
For the Speakers
I have one request for the speakers who are typically excellent. When you refer in a talk to someone else’s work don’t just use their first name. More than once I heard a speaker mention Paul and Kimberly or Adam or Patrick and I guarantee you that half of the people in the session don’t know who they are. Instead, introduce people properly by citing their full name and credentials, it’s more professional and doesn’t lose people.
I’ll be going to Florida next week to attend SQL Rally and present a session titled The Data Mining Lifecycle. I’ve spoken on data mining at least a half dozen times at different venues in the last year and I always get a good crowd. I’ve never taken a poll of the audience to find out why they decided to attend, but I think I have a good idea.
Why should you attend this session? You might be curious about data mining since you’ve seen it in SQL Server but you haven’t had a chance to dig into it. You’ve read about the future of in-database predictive analytics and want to know more. You learned about the mathematical part of data mining algorithms in college but haven’t had a real problem to apply them. Or you want to earn bonus points at your company by improving the ROI on a project you’re working on. These are all good reasons to attend, and it probably describes 99% of the people who have attended my sessions in the past.
At SQL Rally I’ll talk about how to get started with data mining by developing a problem statement and setting a target. These problems are all around us and yearning to be solved. As DBAs and developers we have data at out fingertips and we work on projects that provide us with great insight into the needs of the business. We hear data mining type questions constantly. How many times have you heard someone in your organization say “I wish I knew how the services we offer impact the customer experience” or “If I knew a customer’s annual income I would be more likely to market them the right products”. This is where predictive methods can provide a big bang for the buck. With SQL Server and the data mining algorithms you could predict a customer’s income, and you can forecast the customer experience based on the services your company provides them.
I’m currently working on a couple of projects from which I’ll use some real-life examples. One project is in the area of foster care and it’s a place where data mining can do so much good, both for the welfare of a child and in spending our tax money more effectively. When a child is placed into foster care, how much time will they stay in care until they are able to leave the system? When they leave foster care, how likely are they to return? What is the likelihood of a foster care placement being a successful placement, where the child is in a safe and stable situation? How likely is a child to be abused while in a particular foster care placement?
I’ll have examples like these and more, at least as many as I can cram into one hour. I hope you’ll attend and that you leave my session ready to make an impact with data mining at your organization.