The use of a MLP neural network for analysis and aodeling of land use changes with variations variable of physical and economic social

Abstract

Modeling the land use changes is a method that used to understand the causes and effects of dynamic changes. The model in this research is the ANN model with Multi-layer Perceptron (MLP) network architecture and backpropagation algorithm. The Artificial Neural Network (ANN) method is a potential method for land change as well as test the predictive abilities that will be produced by the model. Land use change modeling uses a combination of ANN and GIS methods. The aim of this research are (1) predict land use and land use change in Banyumanik District in 2011, 2015 and 2019, (2) build the land use model using the ANN method and (3) predict of land use in Banyumanik District in 2027. To predict the land use change is use Markov Chain models. The purpose of modeling land changes is as well as the factors that drive these changes. Some of the drivers of land use changes are physical and socio-economic variables. Physical variables are distance to road, distance to river, distance to agricultural and vacant land, elevation, slope, and climate. Whereas the economic variables are population density, and market land prices. Physical data variables obtained from high resolution satellite image processing. For socio-economic variables data are obtained from statistical data and field surveys. In this research, the model is carried out in a framework with various variables that are different, so that the best model is obtained. Cramer’s V value each variable is tested to see the relationship between these variables.