Application of generalized additive model location, scale and shape (GAMLSS) for rice production in Banyuwangi regency

Abstract

Banyuwangi is one of the regencies in East Java that has great potential for rice production. In terms of rice production, Banyuwangi occupies the top 5 positions in all East Java districts.The focus of this research is to get a suitable model for rice production and find out how the influence of the predictor variables on the response variable. The models used in this case is Generalized Additive Model Location, Scale and Shape (GAMLSS) with RS algorithm.GAMLSS is an extension of the GAM and GLM. GAMLSS is also a semi-parametric which has advantages, namely (1) the distribution used is wider, because it contains the distribution of the exponential family and other distributions and (2) all of the parameters (μ, σ, υ, τ) can be modeled depending on the distribution used. The analyzed model contains rice productionas a response variable and predictor variables used: total rain fall, harvested area, population in district, students ofsenior high schools, and health centers.In this study, the best model will bedefinedby the smallest AIC and SIC from comparison results of all possibilities in vary smoothings, parameters, and degrees of freedom. Furthermore, the best distribution for rice production will also known in this study.