Determining the weights of influencing factors of construction lands with a neural network algorithm: a case study based on Ya’an City

• Published in: Earth science informatics Vol. 14; no. 4; pp. 1973 - 1985
• Main Authors: Gao, Lei, Zhou, Yazhou, Guo, Kairui, Huang, Yong, Zhu, Xiaofan
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.12.2021; Springer Nature B.V
• Subjects: Algorithms; Analytic hierarchy process; Artificial neural networks; Back propagation networks; Case studies; Coefficient of variation; Construction; Correlation coefficient; Correlation coefficients; Earth and Environmental Science; Earth Sciences; Earth System Sciences; Information Systems Applications (incl.Internet); Mathematical models; Neural networks; Ontology; Research Article; Simulation and Modeling; Space Exploration and Astronautics; Space Sciences (including Extraterrestrial Physics; Area forecast; Neural network; Weight determination; Machine learning
• ISSN: 1865-0473; 1865-0481
• DOI: 10.1007/s12145-021-00657-8

In-depth quantitative studies regarding the nonlinear development laws of a region and the calculation of construction land area (CLA) and weight coefficients of different influencing factors are of great significance for establishing the developmental change model of a region. In this study, a novel calculation method using correlation coefficients was introduced. First, models of three neural network algorithms, namely, back propagation neural network (BPNN), grey model neural network (GMNN), and generalised regression neural network (GRNN), were constructed. Key attention was given to improvement of the BPNN. The correlation laws of different influencing factors in a region and the CLA were discovered and extracted using these three models. The coefficient of determination and coefficient of variation were applied to verify the validity of the simulation results for each neutral network algorithm model. The mean absolute error and root mean square error of the three algorithm models were calculated to select the neutral network algorithm model with the highest accuracy. Subsequently, the mean impact value algorithm was added to the selected algorithm model to calculate the weight coefficients for the different relevant factors. The calculated results were compared with the weight values of influencing factors, which were calculated using the analytic hierarchy process, thus forming a set of calculation methods for a more accurate judgement of the influencing factors and weight coefficients of CLA. In this study, the correlations between the influencing factors and CLA were calculated using the proposed method via a case study. The calculated results of the proposed method conformed well to the practical situation in Ya’an City, indicating that the proposed method is worthy of promotion and practice.