 |
Nate Silver put big data on the map when he projected Obama’s presidential victory with such precision in 2008. Silver successfully predicted the outcomes in 49 of the 50 states in the 2008 U.S. Presidential election. And again in the 2012 United States presidential election between Barack Obama and Mitt Romney, Silver correctly predicted the winner of all 50 states and the District of Columbia. That seems pretty amazing given the wide variability in the projected results. Just look at the current projected Iowa caucus winners for the Republican nomination. As of the writing of this blog (Wednesday, January 27, 2016), the projected winner ranges from Ted Cruz by +4 to Donald Trump by +11 (see Figure 1).  Figure 1: Recent Republican Polling Data The PredictIt marketplace, which is modeled after the famous Iowa Electronic Marketplace (IEM), allows you to actually buy shares in candidates creating a stock market for politics. The underlying principal behind PredictIt and the IEM is that you can get more accurate and honest data when people have their own money invested. People may lie with their opinions, but when it’s time to put your money where your mouth is, you get a more accurate answer. The PredictIt marketplace also projects Donald Trump as the winner (see Figure 2).  Figure 2: PredictIT Iowa Caucus Projected Winner However, Nate Silver’s FiveThirtyEight site is projecting Ted Cruz as the winner. And given Nate Silver’s track record, it’s worth our time to see how his methodology differs from the other two (see Figure 3).  Figure 3: FiveThirtyEight Republican Iowa Projected Winner FiveThirtyEight collects data from a number of different polls, and since not all polls are created equal, the FiveThirtyEight forecasts are calculated based on weighted polling averages. The weights account for the quality of each poll based on its track record and its methodological standards. They also account for sample size and how recently it was conducted; recent polls are weighted much more heavily than older ones. Polls are also adjusted for skewing where skewing is the tendency for certain polls to consistently favor certain candidates more than the average of other polls. For the Iowa Republican caucuses, FiveThirtyEight has collected 76 polls. Below is a sample, ranked by how heavily they factor in to our latest polling averages (see Figure 4).  Figure 4: FiveThirtyEight Presidential “Election Score” By golly, this looks like a FICO score where the FICO scores gathers and weighs multiple variables to determine the likelihood that someone will repay a loan. In the case of the Presidential nomination, FiveThirtyEight has created an “Election Score” that weights differently the data from 76 different polls. Role of ScoresIn our “Thinking like a Data Scientist” 8-step process (covered in my blog “Thinking Like A Data Scientist Part III: The Role of Scores”), I discussed how scores are an important and actionable concept for business stakeholders who are trying to envision where and how data science can improve their decision-making in support of their key business initiatives. The beauty of a “score” is its ability to integrate a wide range of variables and metrics into a single number, and the power of the “score” is the ability to start small and then constantly looking for new metrics and variables that might yield better predictors of performance. Simple but powerful, exactly what big data and data science should strive to be. To learn more about EMC’s unique approach to leveraging Big Data to drive business value, please check out EMC’s Big Data Vision Workshop offering. |
