Data mining allows the computer to lead the way to look for interesting findings. It is more exploratory than other forms of analytics. Sometimes “begin with the end in mind” is not the right approach for analytics. This approach is about letting computers do what that are good at, which is mining massive data sets. You can think of data mining as “letting the data speak.”

Data mining is especially helpful for raw, unstructured data. Examples of this are text or audio files.

Data mining can discover relationships in the data that may not have occurred to a human user of the data. It can also identify outliers in the data that may be of most interest to an end-user.

One caveat is that data mining can highlight “spurious correlation.” Sometimes patterns show up in the data that do not reflect any clear underlying rationale. One risk of data mining is the term “digging” as in “you’re digging for answers.” With big data, you can usually find some sort of pattern, but that doesn’t mean that it is legitimate or sensible.