Big Data Mining: Classifier Performance Analysis

Data mining is a vast field that combines different areas like artificial intelligence, database management, data visualization, machine learning, mathematical algorithms, and statistical techniques.Big data is about large amounts of data coming from a variety of sources such as social networks, sensors, devices, third parties, web applications and social media and in a variety of formats such as text, video, audio, diagrams, images, and combinations of two or more formats.Performing data mining tasks on large-scale data sets, Big Data, is called Big Data Mining, and is considered a complex activity.To perform big data mining tasks, data scales far beyond the capacity of a single computer.The big data mining platform needs to rely on clustered computers on high-performance computing platform.This study aimed to perform a performance evaluation of complex classifiers in a local environment and in a cloud environment, using data mining and distributed processing tools.The methodology used for this research was experimental.The data set used in the experiment were related to Fatec students from the Sorocaba and Indaiatuba campuses, the cloud computing platform used was Microsoft's Azure.5 virtual machines were created to build the cluster.Through this study it was possible to observe that there was an average drop of 66% in the processing time of the classifiers in the cloud in relation to the local machine, it was also possible to stipulate an ideal virtual machine configuration for the analyzed dataset, thus avoiding wasted resources.