PREDICTION OF STUDENTS DROP OUT WITH SUPPORT VECTOR MACHINE ALGORITHM

Abstract

The quality of a university can be seen from the high level of student success and the low level of student failure. As for the cause of student failure is the case of drop out. To overcome these problems, predictions are made using the support vector machine method. The Support Vector Machine tries to find the optimal hyperplane where the two pattern classes can be separated maximally, the parameters used in the Support Vector Machine are only kernel parameters in one C parameter which gives a penalty on randomly classified data points. In the Support Vector Machine the weights (w) and biases (b) are global optium solutions from quadratic programming so that just running once will result in a solution that will always be the same for the same kernel and parameter choices. Through the implementation of the support vector machine, it is expected to get the parameters of the Support Vector Machine that are used correctly to obtain the best margin in predicting students dropping out.