Investigation of regression in the data mining process.

Author(s)

Daniel Gomes Dosualdo

• Document type: Master's Dissertation
• Press: São Carlos.
• Institution: Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC/SB)
• Defense date: 2003-05-30

Examining board members

Solange Oliveira Rezende; José Augusto Baranauskas; Juan Manuel Adán Coello

• Advisor: Solange Oliveira Rezende

Abstract

Data mining refers to the process which are able to find patterns from big amounts of data in order to discover knowledge. After found the patterns, the post-processing stage of Data Mining evaluates some aspects of these patterns such as precision, compreensibility and interessability. The activity of regression in Data Mining tries to predict the values of a continuous target variable based on a set of other variables. Beside the fact of many researches in Machine Learning and Data Mining are concerned to classification problems, there are many real world regression problems. This fact motivates the study of methods related to post-processing in symbolic regression. Moreover, a group of researchers of Computational Intelligence Laboratoiy (LABIC) is developing a research project, called DISCOVER. The aim of DISCOVER is to support the stages of knowledge discovery offering methods to Machine Learning, Data Mining and Text Mining. This work, that is related to regression problems, is one of the projects integrated into the DISCOVER. In this work we proposed and implemented a computational environment, the DISCOVER POST-PROCESSING ENVIRONMENT OF REGRESSION - DiPER - which is a framework implemented according the specifications of DISCOVER project, that offers a collection of methods to be used in the post-processing stage of Data Mining. (AU)