Prediction of research performance by academicians in local university using data mining approach

摘要

The performance of academicians can be evaluated based on three aspects, which are teaching activities, research activities and community service. This study focuses on the performance of research activities by academicians in UiTM Shah Alam. Research performance of academicians is measured based on the value of the Hirsch index (h-index). Currently, UiTM does not have a specific and statistical method to predict the performance of academicians based on research activities. Therefore, the aim of this study is to predict the research performance of the academicians. Data mining is one of the common approaches used in solving problems that arise in the educational sector. Interesting information can be extracted from the data. The data for this study were obtained from Institute Research Management and Innovation (IRMI). The data were analysed to determine the suitable data mining approach for the research performance and to identify the factors that influence the research performance. In this study, four different types of classification models were used, which are Logistic Regression, Decision Tree, Artificial Neural Network, and Support Vector Machine by using SAS Enterprise Miner software. The performance of each model is estimated by accuracy, precision, sensitivity and specificity performance metric. The finding in this study indicates that the Decision Tree using Gini splitting criteria is the best model with the highest accuracy of 84.17%. The factors influencing the research performances are the total number of published article, the total number of attended conferences and age of the academicians.