Forecasting Saffron Yield Using Data Mining and Determining Climatic Parameters Influencing its Yield in the Province of Khorasan Razavi

Saffron is the world's most expensive plant and plays a significiant role in economic and social conditions in arid and semi-arid areas, thus annual yield prediction of saffron is very important. In this study, the prediction of saffron annual yield and identification of suitable areas of culture based on climatic parameters and climatic parameters affecting the saffron yield using data mining in the province of Khorasan Razavi is introduced. The results showed that by increasing the average temperature of the cold months and decreasing the average temperature of warmer months, the crop yield increased. Increased rainfall and sunshine hours during the growing season resulted in enhanced yield. Saffron yield had the greatest negative correlation with maximum temperature of April, the lowest negative correlation with the maximum temperature in May and the greatest positive correlation with maximum temperature of February and the lowest correlation with the minimum winter temperature. The results showed that various data mining techniques predicted suitable areas for cultivation better than annual yield. Among the models used, linear support vector machine (SVM-linear), radial basis functions SVM (SVM-rbf) and classification discriminant models had a better performance than others and their determination coefficient were 0.95, 0.93 and 0.84, respectively. Moreover, a good correlation was obtained between the area of cultivation and the saffron yield.