Scholarlink

<p>Accurate yield forecasting is crucial for optimizing agricultural practices and ensuring food security. In this study, we propose a novel Successive Integration Hybrid Forecasting Model (SIHFM) that combines statistical predictions with dynamic modelling to forecast the yield of cotton crops. Utilizing historical yield data spanning from 1964 to 2020, we used autoregressive integrated moving average (ARIMA), long short term memory (LSTM), and gated recurrent unit (GRU) models to generate both statistical and dynamic predictions. Initially, statistical predictions were made based on the yield of preceding years, followed by dynamically predicting subsequent year’s yields using these statistical forecasts as inputs. This hybrid methodology offers a unique perspective by leveraging the inherent patterns captured by statistical models to inform dynamic forecasting using our proposed model SIHFM, thus enhancing predictive accuracy. Furthermore, deep learning methods such as LSTM and GRU were utilized to capture complex temporal dependencies and nonlinear dynamics, further improving the predictive capabilities of the model. Evaluation metrics including Root Mean Square Error (RMSE), Correlation Coefficient (CC), Mean Squared Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) are employed to assess the performance of both approaches. Our proposed model i.e. SIHFM using LSTM and GRU demonstrated promising results, showcasing its effectiveness in accurately predicting cotton crop yields. This innovative framework provides valuable insights into harnessing the predictive power of statistical models to enhance dynamic forecasting in agricultural yield prediction, offering promising implications for precision agriculture and decision-making in crop management strategies.</p>

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

A novel hybrid time series deep learning model for forecasting of cotton yield in India

  • 期刊论文

  • 发表日期：

    2025年1月15日

查看全文
  • Neetu Agarwal,
  • Neelu Choudhry,
  • K C Tripathi

摘要

Accurate yield forecasting is crucial for optimizing agricultural practices and ensuring food security. In this study, we propose a novel Successive Integration Hybrid Forecasting Model (SIHFM) that combines statistical predictions with dynamic modelling to forecast the yield of cotton crops. Utilizing historical yield data spanning from 1964 to 2020, we used autoregressive integrated moving average (ARIMA), long short term memory (LSTM), and gated recurrent unit (GRU) models to generate both statistical and dynamic predictions. Initially, statistical predictions were made based on the yield of preceding years, followed by dynamically predicting subsequent year’s yields using these statistical forecasts as inputs. This hybrid methodology offers a unique perspective by leveraging the inherent patterns captured by statistical models to inform dynamic forecasting using our proposed model SIHFM, thus enhancing predictive accuracy. Furthermore, deep learning methods such as LSTM and GRU were utilized to capture complex temporal dependencies and nonlinear dynamics, further improving the predictive capabilities of the model. Evaluation metrics including Root Mean Square Error (RMSE), Correlation Coefficient (CC), Mean Squared Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) are employed to assess the performance of both approaches. Our proposed model i.e. SIHFM using LSTM and GRU demonstrated promising results, showcasing its effectiveness in accurately predicting cotton crop yields. This innovative framework provides valuable insights into harnessing the predictive power of statistical models to enhance dynamic forecasting in agricultural yield prediction, offering promising implications for precision agriculture and decision-making in crop management strategies.