APLIKASI REGRESI LOGISTIK ORDINAL MULTILEVEL UNTUK PEMODELAN DAN KLASIFIKASI HURUF MUTU MATA KULIAH METODE STATISTIKA

Abstract

Statistical Methods  (STK211) is an interdept course under coordination of  Statistic Departement Faculty of  Mathematics and Natural Science, Bogor Agricultural University (BAU). The final grade received by student  who follow Statistical Methods is measurement  in ordinal scale,  that is A, B, C, D and E.  In the 2008/2009 academic  year  there  are  7  parallel  classes  in  the  Faculty of  Mathematics  and  Natural  Science,  BAU.  By considering the hierarchical structure contained  in the score of student achievement data, the student (first level) is  nested in a parallel class (second level), hence this study used multilevel ordinal logistic regression analysis  to  model  the  final  score  of  Statistical  Methods  with  the  factors  that  influence  it.  Explanatory variables that significantly affect the final score of Statistical Methods are the GPA of TPB (student’s first year of college) and gender, with the variability of the intercepts across parallel classes in the logit function as 1.184. Percentage classification accuracy obtained by using multilevel ordinal logistic regression model was 56.85%

Keywords : hierarchical, multilevel modeling, multilevel ordinal logistic regression, classification