Using Clustering Algorithms of Machine Learning for the Economic Assessment of Land Consolidation Projects

Abstract

This study evaluates the economic impact of land consolidation by predicting profitability changes using machine learning techniques. Research was conducted in the Kızılcabölük neighborhood of Denizli, Turkey, using field-based data on parcel structure and farm inputs. Several algorithms - artificial neural networks, decision trees, and linear regression - were tested. Linear regression achieved the best performance (RMSE: 0.0043 validation, 0.0031 testing). Sensitivity analysis showed parcel area, parcel number, and labour as the most influential variables. The results demonstrate that machine learning can reliably estimate post-consolidation profitability using only pre-consolidation data, providing a practical decision-support tool for land consolidation planning.