MODELLING MATHEMATICS LEARNING OUTCOMES USING A MULTIPREDICTOR SEMIPARAMETRIC REGRESSION APPROACH BASED ON SPLINE ESTIMATOR

Education is one of the points of Indonesia's SDGs which is stated in goal number 4. Mathematics is one of the subjects that contributes to the realizing national education goals. In the independent curriculum, the success of the learning process at school can be seen from the criteria for achieving learning objectives. In this article, we analyzed students’ mathematics learning outcomes using a multi predictor semiparametric regression approach and interpreted the results with Spline estimator. The results shows that the differences between the types of classes greatly influence outcomes in learning mathematics, where social classes experienced a decrease of 2.435 percent compared to science classes. To increase outcomes in learning mathematics, the percentage of learning motivation must be more than 88 percent. Apart from that, high or low IQ cannot determine whether students’ mathematics learning outcomes. Furthermore, by combining linear and nonlinear components in the model effectively, the overall accuracy based on the MAPE value is 7.87 percent, so that the model can be predict the actual value high accurately. Thus, the multi predictor semiparametric regression approach based on spline estimator can explain the mathematics learning outcomes model very well.