A pattern makes the difference between data mining on the one hand and simpler statistics on the other.

Whereas the latter usually just AVERAGES ALL data sets, a data mining tool detects PATTERNS which are:
* significantly DIFFERENT from each other
* comprising within one single pattern only data that is very SIMILAR

Example: there is a big birthday and the particpants are analysed. The statistics tool determines that the average age of particpants is 40, the average salary 1500 Euros, gender pretty insignificant. Instead, data mining would e.g. find 3 different patterns: one with average age 67, salary (penion) 1200 Euros, and more females (higher mortality for men in this age group); a second pattern would show average age of 42, salary of 3800 Euros, gender insignificant; and the third pattern has an age of 11. Hence, data mining patterns are by and large much more revealing.

Note: good mining tools work with multi-variate models, and they also determine the most important/ correlated drivers for each pattern.
Source(s):
www.EasyDataMining.com

Permalink | Report