S3 Home‎ > ‎Our Work‎ > ‎Areas of Expertise‎ > ‎


Analytics is defined as the discovery of interesting patterns in data. We use these patterns to answer questions: "Are Kareem Abdul Jabbar and Henry VIII similar people?" The answer depends on how we define the pattern. Kareem and Henry were both extremely tall for their time, and both are known as avid collectors of oriental rugs. Measured by these patterns, the two are quite similar. But if we adjust the pattern just a bit, the similarity disappears: while Henry, at 6'2" was very tall for his day, his absolute height is substantially less than Kareem's at 7'2", and where Henry's rug collection focused on Turkish pieces, Kareem's is mostly Persian. 

The field of analytics deals with ways of adjusting and defining these patterns and more importantly, ways of discovering these patterns in data-driven pipelines.Typically, some pre-processing (data fusion) occurs before we can do effective analytics. This is because the "things" we find patterns in are usually not represented by the raw data, and a common mistake is to seek patterns at too low a level of abstraction. In the example of Kareem and King Henry, the raw data will likely include many observations of both (and of other people, too). So the raw data lacks sufficient structure to see the patterns we are interested in. When we fuse the observations of Kareem into a model for him, and fuse the observations of King Henry into a model for him, we have sufficient structure to discover the patterns of interest. 

Said differently, when we fuse the raw data we get situational awareness. When we apply analytics to that awareness we get situational understanding.

S3 Data Science, copyright 2015