Model Base On Human Resource System Using Classification Technique

In higher education such as university, academic is becoming major asset. The performance of academic has become a yardstick of university performance. Therefore it's important to know the talent of academicians in their university, so that the management can plan for enhancing the academic talent using human resource data. Therefore, this research aims to develop an academic talent model using data mining based on several related human resource systems. The case study used 7 human resources systems in one of government university in Malaysia. This study shows how automated human talent data mart is developed to get the most important attribute's of academic talent from 15 different tables like demographic data, publications, supervision, conferences, research, and others. Apart from the talent attribute collected, the forecasting talent academician model developed using the classification technique involving 14 classification algorithm in the experiment for example J48, Random Forest, BayesNet, Multilayer perceptron, JRip and others. Several experiments are conducted to get the most highest accuracy by applying discretization process, dividing the data set in the different interval year (1,2,3,4, no interval) and also changing the number of classes from 24 to 6 and 4. The best model is obtained 87.47% accuracy using data set interval 4 years and 4 classes with J48 algorithm