Multivariate Adaptive Regression Spline (MARS) Model On Dengue Hemorrhagic Fever (DHF) Sufferers In Semarang

Abstract

Regression model examines a functional relationship between response variables and predictor variables. If data are non-patterned and no a priori model information about the retrogression curve is available, a nonparametric regression model called Multivariate Adaptive Regression Spline (MARS) can be used. MARS is the combination between spline and Recursive Partitioning Regression (RPR), and therefore it can yield continuous estimation of regression functions. Three components constructing MARS model include basis function, knot, and interaction. Recursive partitioning approximates an unknown function using a developed basis function. Dengue Hemorrhagic Fever (DHF) is one of health problems of which incidence shows an increase year after year. DHF analysis is carried out on survival period and can be modeled using MARS. Survival period is defined as individual’s probability function to survive in certain time; in this case, individual is fully recovered. The present research aims at finding out a model of DHF sufferers’ survival period using MARS and its influencing factors. Data of the research include the 2013 medical record data obtained from Semarang Department of Health. The research results in survival period model (MARS) as well as its influencing factors, such as age (  ), sex (  ), trombocyte (  ), hemoglobin (  ), hematocrit levels (  ), and immunologic response (  ).