6 Things We Learned about Enterprise Analytics from Nate Silver
Here are half-dozen quotes and comments that Silver made during his Gartner keynote that are relevant to the enterprise analytics space.
When it comes to the trendy topic of big data, Silver pointed to a few major problems at the very start. For one, Silver warned about mistaking a bug or anomaly for an opportunity, citing some of the odd recommendations from readily used GPS and online mapping systems. In addition, many business number crunchers are desperately seeking a signal through the noise. Now, with more and more data available, businesses are given more opportunity to cherry pick the results they want to see.
Business leaders often become enamored with the Moneyball effect, thinking the manner in which analytics and stats were used in professional baseball can be directly applied to maximize enterprise data. The correlation isnt that clear, said Silver. In baseball, winning and losing can be attributed to individual players, and that doesnt translate well with day-in, day-out business and project teams. Silver recommended an internal definition of progress based on what youre measuring for.
Not mired in analytic gems or woe, Silver fleshed out a few direct suggestions for real-world analytic programs and practitioners. First, think probablistically; expect numbers to guide but not universally solve all lingering business questions. Secondly, Silver emphasized that businesses know where youre coming from. He noted the practicality of data models to remove some obviously ill-fitting candidates from the massive flow of applicants to a universitys degree program.
Another suggestion Silver passed along in to tackle big data was trial and error, or the idea that complex problems and programs usually involve a learning curve. Just starting with advanced analytics will help your practitioners move past theoretical discussions and more clearly into business realities with data. Even though this may be painstaking, you still have to do it, Silver said.
People have a small, subjective viewpoint in the big world out there, Silver said. This can be particularly damaging, but not always readily evident to the business leaders themselves or in analytic outcomes. Not being shy about bias leads to better business questions from the start as well as quality outcomes. Its also important not to create a cloud of bias based on returns in the present, which has shown to be a disastrous issue in recent global financial turmoil.
Silver said he primarily programs in Stata for models after years of becoming the MacGyver of Excel but years of work have only confirmed for him that goals should be the central concern of data programs. Tools are important and efficient code is important, but at the same time, the attitudes you adopt toward this and a solid understanding of what your goals are ... those are more fundamental issues than which software package youre using.
Check out more on the Gartner BI event here.
For Silvers daily political and statistical analysis, read his blog at The New York Times.
To sign up for our weekly analytics newsletter, register here.
Nate Silver image courtesy of Wikimedia Commons. All stock images used with permission from Thinkstock.
To a packed keynote audience at Gartners BI event in Grapevine, Texas, celebrated political prognosticator Nate Silver put his wide analytic expertise on display. Along with touching on the way Bayes and data models can and cant impact natural disaster forecasts and chess mastery, Silver dropped plenty of tidbits about the practical strategy and application of analytics.
5 Mistakes That Can Kill Analytics Projects
SEPT. TOP READER PICK What the Top 14 Relational Database Skills Pay
SEPT. TOP READER PICK What the Top 9 Data Visualization Skills Pay
SEPT. TOP READER PICK What the Top 15 Machine Learning / Statistics Skills Pay
SEPT. TOP READER PICK What the Top 15 Data Management / Big Data Skills Pay
7 Key Considerations When Choosing a Data Pipeline Service
8 Steps to Success With Self-Service Analytics
What the Top 8 Hadoop-Related Data Skills Pay