Mineração de dados distribuída e escalável usando Apache Mahout

Abstract

Huge data sets have been generated from computing tools. Implicit patterns could be present in this data. Data mining worries in look for relationship, specially, in large data sets, enabling the extration of useful new information. Distributed computing allows the data decentralization and speeds up the data mining process. Apache Mahout is a distributed data mining tool, which uses MapReduce program model, promising scalability by spliting the workload in independents tasks, among themselves. This work has as objective to verify Apache Mahout’s performance, through a implemented algoritms’ choice, data set preparation and mining of these data in differents distributed environments, analyzing the tool’s scalability, as the performance improvement due to nodes’ or cores’ addition to the processing.