The science and practice of Predictive Analytics is well-established and rapidly gaining ground in the public and private sectors. It’s not magic anymore because we now have Business Intelligence (BI) systems that harness and organize massive amounts of disparate data and model that data in ways that allow humans to be proactive and make informed decisions. We recently posted a two-minute guide to selecting the right descriptive, predictive, and prescriptive analytics. To review:
|Type of Analytics||What does it do?|
|Descriptive Analytics||Answers the question: “What happened? Using data aggregation and data mining techniques.|
|Predictive Analytics||Answers the question: “What could happen?” using statistical models and forecasting techniques to understand the future.|
|Prescriptive Analytics||Answers the question: “What should we do?” using optimization and simulation algorithms to suggest the best course of action.|
Predictive analytics encompasses a variety of statistical techniques from predictive modeling, machine learning, and data mining that analyze current and historical facts to make predictions about future or otherwise unknown events.
Predictive models exploit patterns found in historical and transactional data to identify risks and opportunities. These models capture relationships among many factors to allow assessment of risk or potential associated with a particular set of conditions, guiding decision making with better accuracy and significant cost savings.
Let’s examine two popular applications: supply chain optimization and crime fighting. How can predictive analytics effectively address these seemingly unrelated topics? Because at a macro level the issues are identical. Consider this abbreviated chronology of our quest to make data-driven (i.e. better) decisions regardless of the setting and the objectives.
- We had no data. We had unstructured observations and gut-feel
- We got some data, but it was incomplete and resided in silos
- We got more (and more comprehensive) data, eliminated silos, filled in the gaps, but lacked modeling tools. This was the era of data-rich but information-poor
Today, predictive analytics tools allow us to compare possible outcomes of events using scenario analysis and foresee challenges and potential disruptions before they happen. Our supply chain optimization use-case comes from a top-10 domestic brewery that used the Halo solution to gain better insight into production. Before Halo, this brewery had plenty of data, but was unable to “beat it into submission and make it tell us something useful about the future.” Sound familiar?
Their data needed to be more easily translated into actionable information for managers and executives. The company had a variety of tools in-house, but the fragmented technical environment was too difficult to manage for quick scalability. They needed a powerful analytics tool to integrate and transform the data from their disparate systems, along with a front-end for visual analytics, designed for the specific challenges of the beverage industry.
A key point of differentiation for Halo was the ability to link multiple data sources to a single visual analytics and reporting package. Halo’s rapid integration framework enables a one-time setup of the BI platform, followed by easy report creation by business users. An early win included creating a daily shipments and depletions report for the CFO. Using a mobile-ready interface, the CFO can quickly scan variances each morning and immediately drill down to SKU and account-level data to see what’s driving exceptions.
Based on these early victories, the brewery believes that the early detection of production efficiencies will yield $550,000 to $800,000 savings within 18 months. In addition, the company points to a 2 full-time equivalent reduction in staffing requirements (about $300,000 annually), and the value of faster decision making by business managers.
Now let’s consider the case of crime fighting and a diagnostic technique called Risk Terrain Management (RTM). The premise of RTM is that location matters. However, it’s no secret that location matters. The question is: how do you utilize data that you currently have to assess spatial risks and prevent undesired outcomes? RTM helps in this process. With a diagnosis of how the environment correlates with certain behaviors or outcomes, you can make very accurate forecasts.
The reason RTM is used by practitioners across many disciplines, not just law enforcement, is because it was originally developed to solve a problem faced by many: how to leverage data and insights from various sources, using readily accessible methods. RTM gained fame as a crime prevention tool, but today it’s being used in urban planning, injury prevention, public health, traffic safety, pollution, and stopping maritime piracy. (Note that this is at its core the same problem the brewery faced, only the vocabulary and the objectives are different.)
In the context of crime prevention, the RTM process begins by selecting and weighting factors that are geographically related to crime incidents. Then a final model is produced that basically ‘paints a picture’ of places where criminal behavior is statistically most likely to occur.
With knowledge of spatial risk factors, intervention activities can be designed to suppress crime in the short-term and mitigate the risk factors at these areas so they are less attractive to criminals for the long-term. For instance, in one National Institute of Justice (NIJ) study, a 42% (statistically significant) reduction in robberies was achieved by focusing on environmental features of high-risk places, not merely the people located there. With RTM, you can prioritize risk factors and prescribe actions to mitigate these factors, even within the confines of limited resources.
As you can see, Predictive Analytics and the underlying tools that support the discipline can be applied in many settings. (We looked at supply chain optimization and crime fighting, but how about the predictive power of Google searches when analyzing the spread of the annual flu bug?) People like to solve problems, but they need the right information. As business leaders we need to make sure they have it and then set them free.