Electronic Theses and Dissertation

Multivariate Adaptive Regression Splines (MARS) merupakan metode regresi nonparametrik yang mampu menangkap hubungan nonlinier antar variabel. Pada penelitian ini, MARS digunakan untuk memodelkan peringkat akreditasi rumah sakit di Provinsi Aceh yang menggunakan data tidak seimbang, di mana jumlah rumah sakit terakreditasi selain paripurna lebih sedikit dibandingkan yang terakreditasi paripurna. Untuk mengatasi ketidakseimbangan data, digunakan metode oversampling, yaitu Synthetic Minority Oversampling Technique (SMOTE) dan Adaptive Synthetic Sampling Approach (ADASYN). Variabel yang dilibatkan meliputi indikator pelayanan rawat inap, seperti Bed Occupancy Rate (BOR), Bed Turn Over (BTO), Average Length of Stay (ALOS), Turn Over Interval (TOI), Gross Death Rate (GDR) dan Net Death Rate (NDR). Hasil penelitian menunjukkan bahwa model MARS untuk data balanced menggunakan SMOTE memberikan kinerja terbaik, dengan nilai GCV terendah (0,0384), nilai R² tertinggi (93,63%), dan APER terendah (3,57%). Selain itu, variabel yang paling berkontribusi terhadap peringkat akreditasi rumah sakit secara berurutan adalah GDR, ALOS, BTO, BOR, TOI, dan NDR.

The Multivariate Adaptive Regression Splines (MARS) is a nonparametric regression method that is able to capture nonlinear relationships between variables. In this study, MARS was used to model hospital accreditation rankings in Aceh Province using unbalanced data, where the number of hospitals accredited other than plenary was less than those accredited plenary. To overcome data imbalance, oversampling methods are used, namely Synthetic Minority Oversampling Technique (SMOTE) and Adaptive Synthetic Sampling Approach (ADASYN). The variables involved include inpatient service indicators, such as Bed Occupancy Rate (BOR), Bed Turn Over (BTO), Average Length of Stay (ALOS), Turn Over Interval (TOI), Gross Death Rate (GDR) and Net Death Rate (NDR). The research results show that the MARS model with data balanced using SMOTE provides the best performance, with the lowest GCV value (0,0384), the highest R² value (93,63%), and the lowest APER (3,57%). Apart from that, the variables that contribute most to a hospital's accreditation status in sequence are GDR, ALOS, BTO, BOR, TOI, and NDR.