Today I had the opportunity to talk to NDU’s graduate students in the Analytics and Simulation for Enterprise Architecture course. We discussed the topic of Enterprise Analytics and Data Science Teams. Key points included the Internet of Things and Data Lakes to bring data and science together.
I appreciate the interest in this growing topic and there were some good questions at the end. Dr. Mark McGibbon had the foresight to include this topic in the syllabus. I look forward to comments on this post. Here is a summary of the discussion points:
Models are only good to the extent that they provide actionable Answers to the Questions that leaders have. -Jay Gendron
- For starters…it all starts with a good question. Give a data scientist a pile of data and they will find something. Will we provide you with something relevant to your business?
- Set the foundations. Big Data (or is it now Fast Data and Different Data?) as well as Analytics
- Data Science is a team sport. Fortunately, it looks like the field will avoid the equivalent of a “webmaster”…analytics is a team sport
- Presented my framework linking IoT-Professionals-Data Lakes. It is a relational thing…up, down and all around
- Internet of Things (IoT). This was a big point of today’s talk
- First, I presented a context based on a just released, market analysis report on the IoT
- Then we had a little fun showing how biologic systems are the future. We enjoyed the first 3 minutes of the talk “Connected Cows?” given by Microsoft VP for Information Management and Machine Learning, Joseph Sirosh, at the Strata + Hadoop 2015 conference
The best air quality monitors we will get is when we can integrate their signal into the internet -Jay Gendron
- Data Lakes. This is not a zero-sum game, and it is a timely and controversial topic. They have both strengths and weaknesses, but to avoid them would appear to be a great loss for data companies…and as Sirosh said in the video “All companies are data companies.”
- After the presentation, there was interest in learning more about two practical analytic mentalities
- Scales of measurements. Understanding the “as is” and “to be” data types in your analytics plan. This topic was solidfied until 1946 in a journal article in Science by S. S. Stevens
- Cleaning and Exploring Data. Both of these are pre-processing steps in analytics and consume over 80 percent of the analytic schedule. Yet, these are two topics that are accessible to a larger audience
Question and Answer Session
The Q&A session was not very long, but two of the more popular questions came up:
- How is this applicable to our space? A: There are many areas of applicability. It does take some good leadership and management combined with creative thinking to tease out the benefits. But, as shown in the Connected Cows? video there are many, many areas yet untapped to apply an IoT mentality. Consider the work already done with text analytics and social media to predict uprise. What could be done with streaming data on weapon performance and health (not unlike manufacturers do with machine spindle speeds today to anticipate failures…and lost production
- How can we afford this Big Data approach? A: How can we afford not to adapt? The technology (such as Hadoop) is already appearing in government organizations. A true cost will be to identify and secure the talent to harness the technology. Consider the ROI in the Connected Cows? video and then look at your own situation to build a business case.
I was very pleased to have the opportunity to present this topic of Enterprise Analytics. What do you think? What portion of the discussion was most most interesting? most concerning? Where do you see Enterprise Analytics going in your organization?
Jay Gendron is a data scientist, business leader, artist, and author who writes about how good questions and compelling visualization make analytics accessible to decision makers. He is an award-winning speaker who has presented internationally. His book Introduction to R for Business Intelligence will be available this summer through Packt Publishing.