Performance evaluation of artificial neural network models in estimate production of yield saffron based on climate parameters

Saffron as the most precise agricultural and pharmaceutical product of the world has a specific place in industrial and export products of Iran. Nowadays, Iran is the largest producer and exporter of saffron in world, as up to 65% of production of this valuable commodity belongs to it. Despite the antiquity of saffron cultivation and added value of this product compare to other current corps of Iran, fewer shares of new technologies is dedicated to saffron and its production is mainly based on indigenous knowledge. In this paper multiple models are conducted in order to evaluate and develop the performance of Artificial Neural Network (ANN) to calculate estimate production of yield Saffron based on Climate Parameters. The calibration and evaluation of models are yielded from the statistics of crop yield and climate factors between years 1988–2007. In order to evaluate models the following statistical criterions are used: Correlation Coefficient (CC), Mean Absolute Error (MAE) and Mean Square Error (MSE). The results are permissible and indicate that the proposed ANN has correlation coefficient of 0.8, MAE of 0.69 and MSE of 0.66 in estimating yield Saffron. Sensitivity analysis of models has shown that the yield production has the most dependency with rain factor, then with temperature factor and finally with humidity factor. Finally, the proposed ANN can enhance the yield Saffron production in climate circumstance of associated area.