| dc.description.abstract | Academically, the success of a student can be demonstrated by Grade Point Average (GPA). The performance of a student can be seen from the academic achievements, potential and motivation from themselves. Success in obtaining a high GPA can not be separated from the factors that affect the intellectual factor (the score of final examination in high school) and non-intellectual factors (gender, enrollment scheme to university, age when enrolled to university, senior high school status, etc). The data used for this research is the secondary data obtained from Student Affairs Directorate Bogor Agricultural University. Random forest method is an ensemble classifier using many decisions tree models. It can be used for classification or regression. This research is aimed to determine the size of the random forest and sample size of the explanatory variables that produces random forest with high prediction accuracy and stability that can identify the most important influential factors in student's academic achievements (GPA). High accuracy and stability can be obtained if size of the explanatory variable (m) and the size of the random forest (k) is determined by the right value. The values of m used in this study are 3, 4, 5 and 6. The values of k used in this study are 100, 500 and 1000. The smallest misclassification rate is at m = 4 and k = 100 with misclassification rate is 39.36%. Optimum random forest trees is used to get the best identifier variables. The most influential variable of the GPA of student at Bogor Agricultural University in the first year study by ranking criteria used Mean Decrease Gini (MDG) value of random forest. | en |
