Business analytics can identify and analyze patters, but perhaps more importantly can reveal the predictive likelihood of an event, and that predictive information can be worth millions and even billions of dollars to companies, organizations, and governments.  In analyzing big data, two trends in analysis have emerged.  First, the typical statistical approach of relying on p-values to establish the significance of a finding is becoming less trusted since with extremely high sample sizes “almost everything” becomes significant.  In contrast, the focus of analysis is shifting more to the size and variance explained, ie examining for example R-squared.  Stepwise regression and cluster analysis are becoming more widely used to supplement traditional p-value analyses.  Secondly in analyzing big data, there is a shift from focusing so much on aggregates or averages to focusing, in addition, on outliers, because outliers oftentimes reveal (predict) critical innovations, trends, disruptions, and revolutions on the horizon.  In essence, knowing more about “who is not your customer and why” may be as (or more) important than knowing about your customer.   Perceptual mapping and multidimensional scaling are being more widely used to explore outlier patterns.

Source:  Based on Gerard George, Martine Haas, and Alex Pentland, “Big Data and Management,” Academy of Management Journal, April 2014, 57, 2, p. 321-326.