Tulīd va Farāvarī-i Maḥṣūlāt-i Zirā̒ī va Bāghī (Feb 2014)
Application of Artificial Neural Networks in Canola Crop Yield Prediction
Abstract
Crop yield prediction has an important role in agricultural policies such as specification of the crop price. Crop yield prediction researches have been based on regression analysis. In this research canola yield was predicted using Artificial Neural Networks (ANN) using 11 crop year climate data (1998-2009) in Gonbad-e-Kavoos region of Golestan province. ANN inputs were mean weekly rainfall, mean weekly temperature, mean weekly relative humidity and mean weekly sun shine hours and ANN output was canola yield (kg/ha). Multi-Layer Perceptron networks (MLP) with Levenberg-Marquardt backpropagation learning algorithm was used for crop yield prediction and Root Mean Square Error (RMSE) and square of the Correlation Coefficient (R2) criterions were used to evaluate the performance of the ANN. The obtained results show that the 13-20-1 network has the lowest RMSE equal to 101.235 and maximum value of R2 equal to 0.997 and is suitable for predicting canola yield with climate factors.