How Predictive Analytics Reveals New Business Insights
Predictive analytics comes in many forms. We show you what each can offer as you look into your highly probable future.
By Jesse Jacobsen, TechnologyAdvice
Predictive analytics' presence as a potent tool for business insight is all but solidified in today's business world. With a variety of business intelligence solutions available to organizations of any size, understanding how to utilize predictive analytics is crucial to gaining a competitive advantage. In a 2012 Accenture survey, only 13 percent of respondents didn't use data insight as at least a minor source of inspiration. Your competition is taking advantage of their data, and it's time you do as well.
Predictive analytics can be very powerful, even bizarre. For example, last year Amazon gained a patent for what they call "anticipatory shipping." Amazon is able to start shipping products to you before you even order them. By utilizing purchase records, website traffic, and geographic relevance data -- among other information -- Amazon starts to ship a product to a region before anyone orders that product and can later specify the exact location following the actual purchase. Once fully implemented, this will lead to even cheaper shipping costs and faster service than their current two-day system, not to mention increased purchase retention because products will reach customers before they have the chance to change their mind.
Although Amazon's new patent sounds like science fiction, it's actually just the result of predictive analytics at its finest. After looking into some of the most common forms of predictive analytics, your organization will also be able to use data to see into the future (or at least the highly probable future).
Descriptive modeling is the simplest form of insight available within predictive analytics systems. Essentially, you use descriptive modeling to assemble data sets into groups based on similar data characteristics.
For example, you can make correlations from data sets to segment your customer base. How many millennial women are purchasing this specific product during this time period? How often do individuals in this region make a purchases in-store versus online? By using this basic analysis, your company can gain actionable market insight. Descriptive modeling alone is useful in many cases but increases in value when combined with the other facets of predictive analytics.
Predictive modeling provides insight into causality and patterns between explanatory and dependent variables. In essence, by looking at the way two related data sets have behaved in the past, you can predict how they will behave in the future.
A common example is credit scores. Your credit score is a simplified way of showing your credit worthiness based on past performance and payment records. Companies can look at your credit score as an indication of how reliable you will be for future payments.
You can use predictive modeling in customer relationship management to determine the likelihood of specific actions taking place. For example, many organizations utilize uplift models to determine the likelihood of a positive outcome after sustained contact through marketing or other services. This is used by companies in the telephone and cable industries when determining the likelihood of customers extending service contracts when their existing contract ends. Health organizations also use uplift models to predict the patient's reaction to a specific action or treatment.
Prescriptive Modeling (often called decision modeling) helps enterprises find the optimal and most likely outcome for a situation. Although predictive modeling tells you what might happen in the future based on data analysis, prescriptive modeling exists to help you determine the best course of action based on your goals, requirements, and limitations.
Prescriptive modeling is especially useful for determining the course of action to take when actual experimentation would be too expensive or risky. For example, some airline companies use prescriptive modeling to determine the optimal flight pricing structures for each flight. These systems take into account a number of factors, such as travel patterns, demand levels, and purchase timing. The resulting ticket prices are designed to optimize profits while generating adequate sales.
UPS utilizes prescriptive modeling to determine the most efficient package delivery routes. Their systems takes into account map data, customer information, and employee work rules (among other information) to optimize their deliveries, reducing travel expenses and late arrivals. All managers and drivers are taught how to input data in the system for unique, optimized routes daily.
Prescriptive modeling lets your organization discover actionable information from your existing data sets. Systems designed to analyze your data in this way can help you optimize pricing, perform critical path analysis, improve resource allocation, and even perform outcome simulation for potential marketing or sales campaigns.
With the variety of models available in predictive analytics, your organization can find a solution that fits your needs. Some enterprises aren't ready for prescriptive modeling, perhaps because they don't see the need or have workable data. Even so, most have enough information to gain competitive insight from descriptive or predictive modeling, especially when combined with data visualization software. You just need to know what you're looking for and how predictive analytics can help you get there.
Jesse Jacobsen is a junior research analyst at TechnologyAdvice. He covers a variety of topics, including business intelligence, project management, and other emerging technology. You can contact the author on Google+.