Many times, the right set of analytical tools to address the needs of the compliance management applications could come from data mining or applied statistics (as well as economics, or operations research). Almost always, in our practice in tackling compliance management problems, we rely on many traditional statistical methods to complement the data mining methods. |  |
|
The lines between data mining or traditional applied statistics are often blurred. The main distinguishing difference in data mining analysis compared to traditional statistical approaches is that it enables data exploration and analysis without any specific hypothesis in mind. Also, while data mining tends to be more application-oriented and looks for useful patterns in a data set, traditional statistical methods are more concerned with modeling and identifying the fundamental nature of the underlying phenomena. Another often cited reason for data mining is that when the data sets get very large and with many variables, rigorous testing of statistical hypothesis on all the potential interesting findings would be overwhelming. The traditional statistical techniques include:  | Exploratory Statistics | | Hypothesis Testing | | Regression (linear, non-linear, logistic) | | Forecasting | | Experimental Design |
|
