Author(s) :

Chris Cheung.

Volume/Issue :

Abstract :

Accurate crop production predictions are crucial for global food security and effective agricultural policymaking. Traditional predictive models often struggle to capture the complex spatiotemporal dynamics qualities in agricultural data. This research aims to improve crop production index predictions by integrating temporal agricultural data from 1990 to 2021 with geospatial information for 162 countries. Advanced time-related architectures, including autoregression (AR), vector autoregression (VAR), spatial temporal autoregressive (STAR), and spatial temporal vector autoregressive (STVAR) models are explored to address the limitations of traditional methods. The study also uses geospatial data to improve the spatial influences inside the models. By combining temporal data (number of years) with geospatial coordinates (longitude and latitude), the research develops predictive models that could better capture the underlying patterns affecting crop production. Various model training measurements are applied to optimize model performance. The outcomes demonstrated that incorporating temporal with spatial data significantly increases the precision of crop yield forecasts as compared to conventional models. The research highlights how the inclusion of both temporal and spatial variables in agricultural predictive modeling can provide useful information for policy makers, farmers and the rest of the actors in agriculture. By creating a platform for using advanced autoregression models and spatiotemporal data integration, it would help improve decision making in agriculture as well as resource management.