PEMODELAN PENYAKIT INFEKSI SALURAN PERNAFASAN AKUT DI DAERAH SEKITAR SEMBURAN LUMPUR LAPINDO SIDOARJO DENGAN PENDEKATAN MODEL MULTIVARIATE ADDITIVE REGRESSION SPLINE

Abstract

The phenomenon of hot mudflow in Sidoarjo is interesting to be investigated further. Regarding the cause, the disaster occurred due to drilling errors resulting in the Lapindo mudflow which resulted in gas emissions causing health problems, especially those related to the respiratory tract, namely respiratory tract infections (ARI). Risk factors that can affect the incidence of ARI in general are socio-demographic, biological, housing and density factors and pollution. Therefore, this study aims to obtain a model for classifying ARI patient data in the Jabon, Tanggulangin, and Porong sub-districts, Sidoarjo district with the variables that contribute to the classification. The nonparametric approach Multivariate Adaptive Regression Spline (MARS) was chosen because several previous studies stated that this method resulted in a higher classification accuracy than other classification methods. In addition, MARS is a classification method that is able to form a model with causal interactions to produce the best MARS model obtained from a combination of Maximum Interaction (MI), Basis Function (BF), and Minimum Observation (MO) values. The results of modeling with MARS there are three variables that contribute to the grouping, namely the percentage of the distance between the house and the source of the Lapindo mudflow, the number of activities outside the house, and the number of house ventilation. The overall model classification accuracy is 97,4 percent with a GCV value of 0,096 and an R² of 82,9 percent