On The Comparison: Random Forest, SMOTE-Bagging, and Bernoulli Mixture to Classify Bidikmisi Dataset in East Java
Nur Iriawan, Kartika Fithriasari, Brodjol Sutijo Suprih Ulama, Wahyuni Suryaningtyas, Sinta Septi Pangastuti, Nita Cahyani, Laila Qadrini · 2018
The Bidikmisi is a scholarship program from the Indonesian government that intended for students who are not economically capable, but they have good academic performance. In the implementation of the Bidikmisi scholarship program, there are indications of a problem, namely the condition of inaccurate allocation in the Bidikmisi scholarship that is accepted or unaccepted. The purpose of this study was to examine several comparison methods that were used to get the accuracy allocation of the Bidikmisi scholarship in East Java. These methods include random forest, SMOTE-Bagging, and Bernoulli mixture model. Based on the AUC and g-mean values, the Bernoulli mixture method has a better proficiency than the random forest and SMOTE-Bagging.