Pemodelan Sebaran Total Dissolved Solid Menggunakan Metode Mixed Geographically Weighted Regression

Abstract

Mixed geographically weighted regression (Mixed GWR) is a technique to explore spatial non-stationarity by considering local and global influences simultaneously. In this study, the Mixed GWR model was used to determine the Total Dissolved Solid (TDS) distribution model on water quality in Pontianak’s. Independent variables used are Chemical Oxygen Demand (COD), Biological Oxygen Demand (BOD), pH, water hardness and color. The results showed that the Mixed GWR model with Fixed Bisquare Kernel weighting was better than the global regression model and GWR because it had the smallest AIC value of 326.48 and MAPE value of 22.34%. Variables that affect globally are hardness and color, while variables that affect locally are COD, BOD and pH.

Modeling the Flood Disaster in South Kalimantan Using Geographically Weighted Regression and Mixed Geographically Weighted Regression

“…In addition, the linear regression model can be used to predict the independent variables [27]. The linear regression model can be written as follows [28]:…”

“…Spatial heterogeneity is detected when the value 𝐵𝑃 > 𝜒 𝛼,𝑝 2 or when the p-value is less than the significance value (𝛼 = 0,05). Next is determining the weighting function representing the observation area data [28]. The weighting function used in the GWR model is the Adaptive Gaussian kernel function, written as follows.…”