Explore New Frontiers in Business Analytics

Converging basic and advanced analytics paves the way for keen insight and business innovation

Big Data Evangelist, IBM

Value judgments are implicit in many technical discussions, as evidenced in the terminology often employed to characterize otherwise prosaic subject matter. I’m as guilty of this result as anybody. One distinction I’ve made for a long time is that of basic analytics versus advanced analytics. The former refers to any business analytics application that incorporates any or all of the following: historical data, decision support, business intelligence (BI), structured reporting, ad hoc query, online analytical processing (OLAP), dashboarding, performance management, and so on. The latter refers to a sprawling range of analytics tools and approaches that have statistical analysis at their core: regression analysis, data mining, predictive modeling, interactive visualization, text analytics, behavioral analytics, graph analysis, stream computing, and so forth. One unfortunate connotation of this basic-versus-advanced distinction is the suggestion that the former is old and boring—but mature and reliable, and that the latter is new and cool—but immature and unproven. However, most analytics technologies listed here—in both buckets—have been around for decades, are substantially mature and reliable, and are continuing to evolve in innovative ways. Consider the diverse IBM portfolio of established, solutions in all these areas1; there is ample evidence that all these approaches have been deployed in prime-time enterprise applications for many years. Another connotation of this distinction is a bit more value-neutral: an implicit assertion about the extent to which each of these technologies has been adopted by businesses and other user organizations. Many adopt basic analytics first and foremost as a core tool of knowledge workers, business analysts, and operational personnel. Adoption of basic analytics is usually through stand-alone tools, applications, and infrastructure geared for these functions. These functions include business intelligence, data marts, performance analytics, operational dashboards, and so on—or through equivalent features that are integral to organizations’ online transaction processing (OLTP) systems. By contrast, to the extent that organizations also adopt advanced analytics, that adoption is often to a more limited extent and among a more specialized class of users—for example, statistical analysts, data mining specialists, predictive modelers, and so on. The people who use advanced analytics tend to fall under the broad heading of data scientists. They have degrees, certifications, and specialized training that enable them to wield these tools and languages to maximum potential.  

The state of advanced analytics

Advanced analytics is, as this category label implies, a catchall for a growing range of approaches, encompassing all the established techniques but also those that are emerging, experimental, and unproven. It is also at the heart of the big data revolution, which focuses on using advanced analytics on trustworthy data to extract otherwise inaccessible insights at ever more extreme scales. In-database analytics—the raison d’être of Apache Hadoop, MapReduce, Hadoop YARN, Apache Spark, and many other big data frameworks—is almost entirely focused on accelerating execution of advanced analytics by leveraging massive parallelism in the cloud and other computing platforms. Advanced analytics is evolving so rapidly and on so many levels that for somebody new to this field to know where to start and where to focus is quite challenging. The Hurwitz Group’s recent study offers an excellent overview of trends in the advanced analytics market.2 The study’s authors, Marcia Kaufman and Daniel Kirsch, discuss how pervasive and embedded advanced analytics has become in the infrastructures, processes, and decision-making culture of many industries. Here is how I interpret the chief trends that they highlight in their report:

  • Workload-optimized, appliance-type systems are the preferred execution platforms for core advanced analytics workloads in many organizations.
  • Cloud platforms are fast gaining adoption for execution and access to the most resource-intensive, next-generation, advanced analytics applications.
  • Vendor-provided, advanced analytics accelerators are gaining traction for speeding value on horizontal and industry use cases.
  • Self-service visualization is making advanced analytics increasingly accessible to knowledge workers.
  • Advanced analytics is infusing big data insights into all business functions, processes, and decisions.
  • Statistical algorithms for advanced analytics are incomplete without visualization tools to deliver insights.
  • The Internet of Things is expected to increase the importance of real-time advanced analytics within stream-computing environments.
  • Adoption of the Predictive Model Markup Language (PMML) enables advanced analytics to be developed and executed in a platform-agnostic manner across disparate data platforms.
  • Open source R is becoming the standard for advanced analytics modeling.
  • The Python language is making advanced analytics accessible to general-purpose programmers.
  • In-database execution can reduce the need for extract-transform-load (ETL) tools to move data between sources and advanced analytics models.

Note that most of these advances—appliance-based deployment, cloud adoption, solution acceleration, self-service tools, democratization of access, advanced visualization, real-time analytics, standards, open source tools, and platforms—are also the focus of continued innovation in the basic analytics arena. In fact, a credible case can be made for the ongoing convergence of the basic and advanced analytics areas. Self-service BI tools, such as IBM® Cognos® Business Intelligence software,3 now integrate advanced visualization and support both with real-time, statistical, and predictive analysis.  

All-encompassing business intelligence

The new generation of BI users is more than run-of-the-mill knowledge workers. Rather, they are power analysts whose tools allow them to do many analyses that previously would have required the help of a professional data scientist. Business analytics is in essence a better term than advanced analytics to encompass all of it, both the established approaches in the two buckets—basic and advanced—plus the new frontiers that are common to both. Please share any thoughts or questions in the comments. 1 Business analytics website. 211 Market Trends in Advanced Analytics,” by Thor Olavsrud, Computerworld, July 2014. 3 IBM Cognos software: Business intelligence and performance management website.   [followbutton username='jameskobielus' count='false' lang='en' theme='light']

[followbutton username='IBMdatamag' count='false' lang='en' theme='light']