The Effect of Soil Physical and Chemical Properties on the Performance Indices of Artichoke’s Leaf using Artificial Neural Network (ANN)

The present study aims to estimate the performance of artichoke via physic-chemical parameters of soil including soil texture, pH, and bulk density using the artificial neural network (ANN) method. Thus, the soils of sixty points across croplands and forests of Golestan province, Iran were sampled, and soil parameters were measured in the lab. Based on the obtained parameters the different models were performed. The experiment was conducted as a randomized complete block design with three replications. The results showed that ANN models were more efficient than the multivariate regression models (MR model). All ANN models were better to estimate plant weight performance compared with the MR model. Plants grown in the soil samples of the “Ahangar Mahalleh area” showed the highest level of yield performance. Based on the findings, model number 5 with a minimum input parameter was selected as an optimal model. All ANN models were better than the multivariate regression models in the estimation of plant weight. As model 5 had almost similar performance with a minimal number of inputs compared with the other models, this model can be selected as the best model.