Building Research Productivity Framework in Higher Education Institution

The purpose of this study is to build a framework for improving research productivity in higher education institutions. The research begins by collecting data and defining candidate variables. The next process is to determine the selected variable from the candidate variable. Variable selection is carried out in three stages, univariate selection, feature importance, and correlation matrix. After the variable selection stage, eight input variables and one target variable were obtained. The eight input variables are Article (C), Conference (CO), Grant (GT), Research Grantee (RG), Rank (R), Degree (D), IPR, and Citation (C). The target variable is Research Productivity (RP). This selected variable is used to build the framework. The next step is to test the framework that has been built. The testing process involves four data mining classifiers, Support Vector Machine, Decision Tree, K-Nearest Neighbor, and Naive Bayes. The classification results are tested using confusion matrix-based testing, accuracy, precision, sensitivity, and f-measure. The testing results show the proposed framework is able to obtain high accuracy scores for each classification algorithm. It means the proposed framework is relevant to use.