Uma metodologia para a seleÃÃo de candidatos ao Mestrado em CiÃncia da ComputaÃÃo do CIn - UFPE usando funcionalidades da mineraÃÃo de dados

AUTOR(ES)

Claudia da Costa Coutinho Salgues

DATA DE PUBLICAÇÃO

2005

RESUMO

The candidates for the Center of Computer Science Postgraduate programme of UFPE, Brazil, are selected based on their CVs, subjectively. Six years ago, as the number of candidates increased, the task of selecting the candidates had to change. Since then, the CVs attributes are weighed as coefficients of linear function variables for producing the candidates scores which are used as the criterion for ranking the students in descending order of classification. The empiricism used for defining the weights is not a precise method, and another important fact is that some good students have not been selected because their peculiar profile have not been captured by the current selection process â the especialistsâ selection process. This work has proposed an expansion of the attributes analyzed, since the current method is limited to CV information, and the construction of a decision support model based on the estimation of the attributes weights from the data. The relevant data for the data mining process was collected â information about 814 candidates to the Postgraduate course in four years, and the performance of the approved andidates â 148 students, was carefully assessed. The framework used for the project development was CRISP-DM - the most widespread methodology describing the steps of the KDD process, for the construction of a classifier more precise and not so restrictive as the current classifier. The most appropriate techniques were chosen for solving the problem, and new models were generated. Logistic regression and linear regression models were compared with the specialistsâ model, and then, that with the best performance was elected. It was proved that better decision support systems than the currently used are viable to be implemented. Other techniques were used for characterizing the students profile suited for the Computer Science Postgraduate programme from UFPE

ASSUNTO(S)

selection process cv information ciencia da computacao computer science postgraduate computer science

Documentos Relacionados

• MineraÃÃo de regras para seleÃÃo de tÃcnicas de agrupamento para dados de expressÃo gÃnica de cÃncer
• Sobre a interferÃncia da produÃÃo cientÃfica e tecnolÃgica da universidade no desenvolvimento local : o caso da CiÃncia da ComputaÃÃo
• Metodologia para desenvolvimento de soluÃÃes em mineraÃÃo de dados: Um estudo prÃtico em diagnÃstico de falhas
• Uma arquitetura de computaÃÃo pervasiva para trabalho de campo
• MineraÃÃo de dados aplicada à cardiologia pediÃtrica