Predictive Analytics for the Modern Enterprise

Chapter 2. Predictive Analytics: An Operational Necessity

Anyone who is part of an enterprise understands that analytics is no longer a sidecar. As we discussed in Chapter 1, the scope of analytics has expanded far beyond weekly and monthly reports or batch processing of multiyear data to build strategic plans. Today analytics is used in everyday applications. When you browse a hotel booking website and it tells you there are only two rooms left in the category you are looking at, this is an example of operational analytics. Another example is when a recommendation engine on an ecommerce website displays photos of other items that people who bought the item you are looking at also bought. Predictive analytics is no exception to this. Analytics is no longer analyzing things after the fact. It is about processing data in flight, learning from it, and in certain cases, providing feedback that the business can use in real time.

The Move from “Data Producing” to “Data Driven”

When it comes to understanding the nature of data, I like to break the topic down into its subcomponents. You’ve likely heard of the term data pipeline, but what does a data pipeline actually entail? Let’s start with where the data originates: in producers.

Most legacy enterprise application data was produced when end users and services interacted with the application or fetched data from other applications using hardcoded integrations—for example, a travel agent booking a flight for a customer on an interface ...