Variable Selection on Indonesian Student Achievement in 2015 Using Regularization of Structural Equation Modelling (SEM)
Eminita, Viarti; Notodiputro, Khairil Anwar; Sartono, Bagus
1 Program Studi Pendidikan Matematika, Universitas Muhammadiyah Jakarta, Jl. KH. Ahmad Dahlan, Tangerang Selatan 15419, Indonesia
2 Departemen Statistika, Institut Pertanian Bogor, Jl. Raya Dramaga, Bogor 16680, Indonesia
Abstract
The Program for International Student Assessment (PISA) is an examination system that evaluates the education system every 3 years to 15-year-old students randomly selected. Indonesia was generally still below the average of the Organisation for Economic Cooperation and Development (OECD). In addition, the results of the PISA study are not only interpreted as rankings, but also important to carry out follow-up based on diagnoses produced from the PISA diagnostic survey to determine the weaknesses and strengths of Indonesian students and evaluate the education system in Indonesia. It is important to consider the use of the right and simple model to see the relationship between the diagnostic results of the performance of students PISA achievements in 2015. The purpose of this study was to select variables that significantly influence PISA achievement in 2015 using the SEM regularization model using The Least Absolute Shrinkage and Selection Operator (LASSO) method. The results of the analysis show that the variables that significantly effects students PISA results are the students insight into the environmental, the students perceptions of science truth, and the students- participation in The International Standard Classification of Education level 0 (ISCED 0).
Keywords: LASSO, OECD, PISA, Science, SEM regularization
Topic: STEM Education