An ensemble deep learning approach for predicting cocoa yield.

Area of Science:

• Agricultural Science
• Computer Science
• Data Science

Background:

• Crop yield prediction is crucial for agricultural planning and policy-making.
• Traditional statistical models for crop yield prediction face challenges like inaccuracy and time inefficiency.
• Deep learning and machine learning offer advanced pattern extraction from large datasets for improved prediction.

Purpose of the Study:

• To propose an efficient deep learning technique for cocoa yield prediction.
• To develop a hybrid Convolutional Neural Network and Recurrent Neural Network (CNN-RNN) model integrated with Long Short Term Memory (LSTM).
• To address the limitations of previous crop yield prediction models.

Main Methods:

• An ensemble deep learning approach using CNN-RNN with LSTM was developed.
• CNN processed the climatic dataset, while RNN handled the cocoa yield dataset for southwest Nigeria.
• LSTM was employed to mitigate vanishing and exploding gradient issues inherent in CNN-RNN models.

Main Results:

• The proposed CNN-RNN with LSTM model was benchmarked against other machine learning algorithms.
• Performance was evaluated using Mean Absolute Error (MAE), Mean Square Error (MSE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE).
• The CNN-RNN with LSTM model achieved the lowest Mean Absolute Error, indicating high prediction efficiency.

Conclusions:

• The CNN-RNN with LSTM model offers an efficient and accurate solution for cocoa yield prediction.
• This deep learning approach overcomes limitations of traditional statistical methods.
• The findings support improved agricultural planning and decision-making through enhanced prediction accuracy.