Data mining based modell aggregation

      In recent years,  data mining based modeling techniques widespread in several sector of the economy. As
competition becomes more and more close now, one of the main competitive edge for market participants is to utilize sophisticated modelling techniques to support CRM and marketing activity.  Furthermore, due to the
New Basel Capital Accord for financial organizations,   there  is a must to develop as sophisticated models
aspossible to avoid the growth of capital adequacy. Thanks to these effects, several models  were developed
topredict customers' behaviour based on  their known attributes. In this paper,  I show that transaction level
models perform better then customer level models, but the interpretation of transaction level models on
customer level might drive to inconsistency. To resolve this problem,  I analyse different methods for
aggregation such as expert method, linear regression and neural network. Finally, I evaluate the efficiency of
different methodologies.
Related Topics:

0 comments:

Post a Comment

 
My Catalyst Copyright © 2011 Designed by My Catalyst