Data Preprocessing in Web Usage Mining
Istrate Mihai · Ovidius University Annals Economic Sciences Series · 2010
Data collected from Web in the first stage of data mining are usually diverse and voluminous. These data must be assembled into a consistent, integrated and comprehensive view, in order to be used for pattern discovery. Like in most applications of data mining, data preprocessing involves removing and filtering redundant and irrelevant data, predicting and filling in missing values, removing noise, transforming and encoding data, as well as resolving any inconsistencies. The task of data transformation and encoding is particularly important for the success of data mining.