CD Tesis
Pemodelan Mixed Geographically Weighted Regression Dan Metode Analitycal Hierarchy Process Pada Kasus Kemiskinan Provinsi Riau
One of the problems in Riau Province is poverty. An important aspect in
overcoming poverty is determining the measurement value. The poverty of an
area and its influencing factor will be different in each region. The difference
can be caused by geographical aspects. Therefore, poverty resolution policies
can also be different in each region so we need a model that can be used to predict
poverty by considering spatial references, one of which is the Geographically
Weighted Regression (GWR) model. There is a problem in the GWR model
that is if some independent variables do not vary spatially, so the GWR model
is developed to be a Mixed Geographycally Weighted Regression (MGWR)
model. Data used in the GWR and MGWR models are poverty rates in Riau
Province (y) with five factors (x). The decision support methods in the form
of recommendations using Analitycal Hierarchy Process (AHP) with estimation
method used is Weighted Least Square (WLS). The weighting function used
is the kernel adaptive exponential. The optimal bandwidth selection uses the
Cross Validation (CV ) method. The best selection criteria used are r2, AIC
and RMSE. The results show that the MGWR model with exponential adaptive
kernel weighting function is better than the GWR model and the AHP method
recommendation areas in the priority of handling poverty are Inhil, Inhu, Kuansing,
Rohil, Meranti, Rohul, Kampar, Bengkalis, Pekanbaru, Pelalawan, Siak,
and Dumai.
Keywords: Adaptive, AHP, Bandwidth, Eksponential, GWR, MGWR
Tidak tersedia versi lain