Spline Truncated Estimator in Multiresponse Semiparametric Regression Model for Computer based National Exam in West Nusa Tenggara

Abstract

Abstract Multiresponse semiparametric regression model is a combination of parametric regression model and nonparametric regression model with response variables more than one and correlate. The estimate used in estimate the parameters is spline truncated. Excess spline truncated is a model that has excellent statistical and visual interpretation and can model data with changing patterns on certain sub-intervals, because spline is a kind of polynomial pieces. The data used in this study is the value of Computer Based National Examination (CBNE) Vocational High School (VHS) in the province of West Nusa Tenggara (NTB) in 2017, each subject tested on CBNE serve as response variables. Based on the significant correlation test results obtained p-value <0.05 so it can be concluded that there is correlation between the responses. The result of the multiresponse semiparametric regression model estimation is obtained by the best model with the value of MSE of 49,608; R2 of 0.84 and minimum GCV value of 0.00000323 so it can be concluded that the value of CBNE VHS in NTB province satisfies goodness of fit criterions.