The paper performs a comparison of data mining tools with conventional tools. It is focused on the analysis of issues such as if the new techniques are better than the older ones if so in which sense and when we should apply them. The effort is concentrated on neural networks automatic interaction detection memory based reasoning and genetic algorithms. These techniques are compared with multiple linear regression and logistic regression using as a standard for comparison a simulated 2 interview magazine customer satisfaction survey.
Author: Thomas Liehr
Company: KANTAR TNS Malaysia
March 17, 2002
- This could also be of interest