Monthly Archives: April 2012

Real-time Business Analytics Delivers Results: Elections

What is the true value of Big Data and Business Analytics?

Better insight and strategy is definitely one answer. But we believe that value needs to be measureable, so we focus on hard numbers – whether that is in increased revenue (from higher value services for example), reduced costs (from better operating efficiency) or higher customer satisfaction (more repeat business, or lower churn).

Some of the best applications for delivering a measureable Return on Investment are the ones where there is a need for rapid, Real-time Analytics, and action based on Big Data is obvious. Some examples are:

  • Network Optimization in Telecommunications
  • Real-time Trade Analytics and Intra-day Risk Management in Financial Services
  • Monitoring and protecting high value assets in Oil and Gas
  • Highly reliable distribution networks in Energy and Utilities
  • Continuous asset scheduling in Transportation Logistics
  • Dynamic Capacity Optimization in Discrete Manufacturing & Supply Chain

But what about some other less obvious uses for Real-time Analytics?

Lately, I’ve been thinking about election results. Though I try to stay out of the fray, I do know that we are in an election year. The pundits and pollsters are off and running with daily, even hourly, predictions as each state checks off another voting day in the Republican primaries. All this talk of numbers and aggregating data on everything from voter demographics to tallying votes come Election Day, is easy to dismiss. After all, primaries are just that—primaries—and according to recent statistics, the very early ones have little bearing on the final outcome and selection of a candidate. Unless of course, you are the one tasked with sorting and analyzing the massive amounts of seemingly unrelated data. Then, it becomes a very fascinating study indeed.

How, you might be thinking, could Real-time Analytics be used to improve elections? After all, politicians never seem to do anything fast. Well, it turns out there are some pretty interesting election requirements. Let’s start with electronic polling. In the 2008 presidential election, more than 130 million people turned out to vote. That is a massive amount of data generated over a short period of time. What happens if a voting machine is unreliable? Or if in the influx of data, statistics from a specific municipality are suddenly “lost,” as happened recently in Florida?

When you have the right Real-time Analytics tools in place, all of the data from reporting precincts nationwide can be aggregated. And thanks to Starview’s one-platform approach, data can be analyzed at a much faster rate than traditional methods that require numerous systems: one to collect and store data, another to analyze it, another to report on it, another that makes a decision and orders an action based on the analysis should any problems arise, and still yet another that triggers that action. With Starview, this all happens within one system.

How would that one system know what to do with highly-specific election data? Well, months before the election, the system would be configured with analytics and rules that are required to identify and respond to issues relating to the election data. In the case of a platform designed to optimize the way votes are tallied and ensure nothing slips through the cracks, engineers could customize the rules to handle scenarios such as: “If a machine in Kansas shows signs that it will fail, then another machine can be delivered to replace it before it fails and the appropriate administrator will also be simultaneously alerted to the change.” This is a highly simplified example, but it gives you an idea of how such rules work. Coding like this in Java can be slow and painful, but Starview’s Enterprise Platform comes with the Star ™ language. Star is a high level language that provides easy to use semantics for processing continuously changing data streams. For engineers tasked with Big Data and Analytics projects, it is very easy to adapt to. Star™ allows for “pattern matching” and “If/Then” rules to be programmed in an easy and simple way. With Starview, it’s possible to take the various brands of voting machines and software that exist nationwide and acquire and transform the data in to a common Active Model that represents election processing world view.

Wait, you say—if it’s not broken, why fix it? Well, as more and more voting locations embrace computer terminals instead of punch-card terminals, more variables will be introduced. With Starview, the platform can be adjusted accordingly and quickly.

As voter information becomes digitized, a one-platform system serves as a safety net. Rules can be customized to weed out voter fraud, such as duplicate voter registrations, people voting on behalf of deceased persons and other scenarios that undermine the efficacy of a democratic system.

Starview’s Enterprise Platform is perfectly aligned to maximize on the many opportunities that are sure to arise in this rapidly changing, data-drenched world. We’ll continue to delve into these possibilities from time to time. In the meantime, feel free to challenge us in the comments section.

We’re asking you: Name a scenario and let’s see if Starview can help solve, or improve it.