Data mining

  toucheatout  2006-03-15 19:26  Data mining  

Principle of data mining

Data mining is the extraction of useful patterns and models among a heap of data. While the analysis and model building phase is made with statistical and machine learning techniques, data mining has to deal with data integration from various sources, data homogeneisation, dealing with different data management systems, addition of percalculated interesting indicators, storage (in datawarehouses) that usually is multidimensional by nature, with aggregation levels (see roll-up and drill-down) on each dimension. The technology of those datawarehouses is often qualified OLAP (On-Line Analytical Processing).

The long and often underestimated process of data gathering and preparation is often what allows for successful mining in the second phase.

Data mining tools

On the free side, you can grab the WEKA data mining tool, a good opensource project integrating a wide variety of models.

A good point-and-click (but priceful) tool was SPSS's Clementine some years ago.