Ferramenta para visualização de diagnóstico de baixo desempenho gerado a partir do método de classificação no processo de mineração de dados, com base nas interações em fóruns de discussão

The tools colaborative and of communication has been used broadly in the education contexts and the Virtual Learning Environments (VLEs), that are a modality of Distance Education (DE), they are more and more being inserted in universities, schools and companies. That communication happens in severa...

Descripción completa

Detalles Bibliográficos
Autor: Silva, Francisco da Conceição
Tipo de recurso: tesis de maestría
Estado:Versión publicada
Fecha de publicación:2015
País:Brasil
Institución:Universidade Estadual do Maranhão (UEMA)
Repositorio:Repositório da Universidade Estadual do Maranhão (UEMA)
Idioma:portugués
OAI Identifier:oai:repositorio.uema.br:123456789/3122
Acceso en línea:https://repositorio.uema.br/jspui/handle/123456789/3122
Access Level:acceso abierto
Palabra clave:Baixo desempenho
Mineração de dados
Classificação
AVA
Fórum
Underperforming
Data mining
Classification
VLE
Forum
Ciência da Computação
Descripción
Sumario:The tools colaborative and of communication has been used broadly in the education contexts and the Virtual Learning Environments (VLEs), that are a modality of Distance Education (DE), they are more and more being inserted in universities, schools and companies. That communication happens in several ways, such as chats, discussion forums, wikis, among others. The forums, especially, consist of spaces for discussions and changes of ideas on defined subjects for their participants, making possible a favorable experience to the learning process. In EAD an recurrent and very challenging problem exists, that is the students' dropout, whose dropout rates are high and preoccupying. In this sense, this research presents the development of a model preditivo of low acting in an AVA, starting from the students' interactions in discussion forums. The objective was to accomplish the forecast of low acting of students, that considered a strong indicator evasion, generating reports that it aids the interested parts in the socket of decision. For that, experiments were accomplished with groups of different data, where the Data Mining (DM) was applied through five classification algorithms, being compared the acting of each one, so that a model with better acting was obtained. For the visualization of the results obtained in the process of DM a tool was developed with the best objective to present the results obtained to the interested parts, being an aid in the socket of decision