AI-driven predictive modelling of national material flows for sustainable resource management and circular economy planning

The paper presents AI-based predictive tools to forecast key material flow indicators at the national level, directly addressing the need for more effective planning in the shift to a circular economy. This study focuses on the Global Material Flows Database for Iraq (۱۹۷۰–۲۰۲۴) and compares linear regression and Long Short-Term Memory (LSTM) models for predicting indicators such as Domestic Extraction, Imports, Exports, and Raw Material Equivalents. Results show that LSTM models significantly lower the mean absolute error, especially for volatile indicators. Although these models inform policy and monitoring, they predict material flow indicators rather than recycling rates; thus, interpretation must account for dataset variability and operational limitations. This research provides actionable insights into resource efficiency and national circulareconomy planning by demonstrating LSTM's superior temporal and structural modeling capabilities.

کلیدواژه ها:

Circular Economy ، Sustainable Resource ، Management ، paper recycling ، Food Waste Recycling ، LSTM ، linear regression ، Material Flow Analysis