Scholarlink

The agricultural production process in the contemporary era is significantly influenced by environmental factors, rapidly changing market conditions, and macroeconomic indicators. Evaluating the wheat production process—a strategic component of agricultural activities—and the factors affecting this process through statistical analysis methods provides valuable insights into production quantity forecasts. In forecasting studies of agricultural data structures in a time series format, both traditional statistical methods and artificial intelligence-based models are widely used. In this study, nonlinear autoregressive with exogenous inputs (NARX) networks, characterized by their dynamic and flexible structure, have been utilized to predict wheat production in Turkiye. Meteorological data (temperature, precipitation), macroeconomic indicators (wheat price, barley price), and other parameters (year, population, wheat cultivation area, wheat yield) associated with past (1950–2021) wheat production quantities have been identified as system inputs. The output variable has been designated as the wheat production quantity for specific periods (1950–2021). The NARX model has been trained using the time series data of these variables, optimized through various training algorithms, and its success in production forecasts has been evaluated using statistical performance metrics. According to the analysis results, they highlight the dominance of temporal, demographic, and economic variables in predicting wheat production, while environmental factors and specific agricultural practices appear to play a minor role in this particular model configuration. This study demonstrates that LSTM-based NARX networks have achieved high accuracy in predicting wheat production levels. The use of such AI-based methods can enable more precise and reliable management of agricultural production processes and offer valuable contributions to policymakers and practitioners in making strategic decisions. Future research could extend these findings by applying similar models to different agricultural products and geographical regions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predictive Modeling of Wheat Production in Turkiye: A NARX Network Approach Incorporating Climatic and Economic Factors

  • 章节

  • 发表日期：

    2025年

查看全文
  • Aslı Akıllı,
  • Mustafa Kan

摘要

The agricultural production process in the contemporary era is significantly influenced by environmental factors, rapidly changing market conditions, and macroeconomic indicators. Evaluating the wheat production process—a strategic component of agricultural activities—and the factors affecting this process through statistical analysis methods provides valuable insights into production quantity forecasts. In forecasting studies of agricultural data structures in a time series format, both traditional statistical methods and artificial intelligence-based models are widely used. In this study, nonlinear autoregressive with exogenous inputs (NARX) networks, characterized by their dynamic and flexible structure, have been utilized to predict wheat production in Turkiye. Meteorological data (temperature, precipitation), macroeconomic indicators (wheat price, barley price), and other parameters (year, population, wheat cultivation area, wheat yield) associated with past (1950–2021) wheat production quantities have been identified as system inputs. The output variable has been designated as the wheat production quantity for specific periods (1950–2021). The NARX model has been trained using the time series data of these variables, optimized through various training algorithms, and its success in production forecasts has been evaluated using statistical performance metrics. According to the analysis results, they highlight the dominance of temporal, demographic, and economic variables in predicting wheat production, while environmental factors and specific agricultural practices appear to play a minor role in this particular model configuration. This study demonstrates that LSTM-based NARX networks have achieved high accuracy in predicting wheat production levels. The use of such AI-based methods can enable more precise and reliable management of agricultural production processes and offer valuable contributions to policymakers and practitioners in making strategic decisions. Future research could extend these findings by applying similar models to different agricultural products and geographical regions.