Hybrid method

In recent years, the fluctuation in agricultural commodity prices in Iran is increased and thus, accurate forecasting of price change is necessary. In this article, a flexible combined method in modeling monthly prices of beef, lamb and chicken from April 2001 to March 2021, was proposed. In this new method, three different approaches namely simple averaging, discounted and shrinkage methods were effectively used to combine the forecasting outputs of three hybrid methods (MLPANN-GA, MLPANN-PSO and MLPANN-ICA) together. In implementation stage of hybrid methods, based on test and error method, the optimal MLPANN structure was found with 3/2/4–6–1 architectures and the controlling parameters are carefully assigned. The results obtained from three hybrid methods indicate that, based on the RMSE statistical index, the MLPANN-ICA method performs the best when forecasting prices for beef, lamb, and chicken. The outputs of three combination approaches show that the shrinkage method, with a parameter value of K=0.25, achieves the highest prediction accuracy when forecasting prices for these three meats. In summary, the proposed method outperforms the other three hybrid methods overall.

Uncertainty and critical conditions have become one of the most important challenges in the world today. In order to achieve superiority in economic competition, the supply chain must be designed in such a way that it has the necessary resilience in critical situations. One of the things that can be considered in this regard is the role of social media in the resilience of the supply chain. Considering the impact that social media has on business processes and considering the rapid and significant advances in information and communication technology, the use of social media in risk situations can be very effective. This research examines the role of social media in the resilience of the digital supply chain with a hybrid approach. The results of this research include 118 open codes, 43 core codes, and 13 main categories. The 13 selected codes of this research are placed in the form of five general categories of causal conditions, contextual conditions, intervening conditions, strategies, and consequences. Causal conditions include 2 selective codes, 8 concepts, and 21 open codes. The contextual conditions include 3 selected codes, 10 concepts, and 25 open codes. Intervening conditions include 2 optional codes, 7 concepts, and 21 open codes. The strategies include 2 selected codes, 7 concepts, and 19 open codes. The results include 4 selected codes, 11 concepts, and 32 open codes.