Association rule‐based multilevel regression pricing and artificial neural networks based land selling price prediction based on market value

Abstract

SummaryIn Tamil Nadu, India, the guideline value (GV) of land for registration typically corresponds to the market price. After registration of land, Tamil Nadu's land registration office determines which areas have GVs that are lower or higher than the government‐fixed value, and depending on the area. Based on determination of price difference, each are GV is raised or lowered. In other countries expect India, different models, including the hedonic price model, quantile, and multivariate regression are used in the current GV fixation. These models have complex relationships with the many different variables that affect the GV of the land. In this article, novel optimized models for fixing land value according to utilitarian land ethics are proposed, including artificial neural networks (ANN) and associative rule‐based multilevel regression pricing models (AMLP). Increase the GV from the current GV in different percentages using the proposed ANN association rule mining method, and then optimize with spatial and infrastructure parameters from the GIS model builder for real‐time GV generation on the website. For real‐time GV generation on the website, the ANN model uses association rule mining and various neural network architectures. Buyer acceptance test validation of the live GV generation reveals 85% accuracy for the proposed model. The proposed models such as ANN and AMLP, prevent problems with GV fixation and boost revenue for the Indian government.