| dc.description.abstract | The common problem in the university is the high
dropout rate. The high dropout rate will have a bad impact on the
university. Various studies have tried to determine the factors that
influence the dropout. Almost all research focuses on academic
factors of students as a determinant of potential dropouts. However,
there are sometimes cases of dropout students who cannot be
determined using academic factors. This raises the hypothesis that
the potential dropout students can be determined from non-academic
factors. There are 5 non-academic factors criteria that can be used
as determinants of dropout, demography, social interaction, finance,
motivation, and personal. These criteria give rise to 37 factors that
are considered influential in determining the potential dropout. The
factors processed into three phases are collecting data, preprocessing
data, and modelling. The factor that are independent to other factors
are the number of family, the interest in the future study, and the
relationship with the lecturer. Based on the result of correlation test
there are two factors had correlation, so the modelling done with two
combination factors. The best model is using combination of factor
the number of family and the relationship with the lecturer using
Decision Tree with split criterion is Maximum Deviance Reduction
and maximum split is 2 with time for training is 1.7386 seconds. | en_US |
