مقایسه انحراف مکانی-زمانی محصولات زراعی تولیدی در ایران بر مبنای الگوی کشت بهینه کشور

Introduction
Nowadays, due to the availability of large amounts of data, data analysis approaches have shown their potential to solve some problems in different economic sectors, for example, the concept of big data analysis has entered various disciplines, such as insurance, banking, agriculture, and environmental studies. Data analysis is performed using one of these three methods, i.e. regression analysis, clustering, and classification. To estimate the relationship between one or more independent variable(s) and a single dependent one, the regression analysis is used; a set of methods that allow the grouping of different agricultural objects is performed by clustering; and classification aims to categorize objects based on their properties, which are called predictors. Some categories of software tools are used for big data analysis such as image processing, machine learning, cloud-based platforms for large-scale information storing, analysis and computation, GIS systems, modeling and simulation, statistical tools, and time-series analysis. The R programming language, an open-source software, is a powerful platform to conduct big data analysis requiring machine learning processes, and statistical operations. This software also acts as a suitable tool for data visualization.
Materials and Methods
The aim of this research is the development of spatio-temporal maps of the value of agricultural crops (in five groups named cereals, legumes, industrial crops, vegetables, and fodder crops) produced in provinces of Iran using the tmap package of the RStudio software. The raw data used in this study was obtained from statistical tables presented by the Ministry of Agriculture Jihad statistics center about the value of different agronomy crops produced in 31 provinces from 2016 to 2020. Furthermore, in this study, some statistical methods were used to compare provinces based on the spatial-temporal deviations of agricultural crops produced from Iran’s official cultivation pattern. The clustering methods utilized herein were the K-means and K-medians methods of the partitioning clustering paradigm, and a hierarchical clustering method.
Results and Discussion
According to the results of this study, large deviations were recorded for Gilan, Alborz, and Yazd provinces followed by Bushehr, Tehran, and Qom provinces. Numerically, average deviations for the three leading provinces were 72.5%, 72%, and 70.95%, respectively. Furthermore, the average absolute deviations of crop yields in the cereals and vegetables categories from the official crop pattern were 5.9%, and 11.7% respectively; while similar measures for the legumes, fodder, and industrial crop categories were 30.2%, 36%, and 41.2% respectively. Moreover, clustering Iran provinces using the K-means and K-medians methods showed that by increasing the number of clusters, the results of these methods converge. Finally, from the practical vantage point, if the clustering curve contains clusters having central symmetry, by exchanging the cultivation patterns of these clusters, the yields of agronomy crops will be changed in the direction of matching the suggestions of the official cultivation pattern.
Conclusion
It is concluded that the existence of reliable input data of agricultural crops produced in provinces, the creation of spatial-temporal maps, and clustering provinces based on the deviations of crops produced in them from the official cultivation pattern helps main decision makers to obtain an appropriate view to establish suitable laws in compliance with matching the real production of agricultural crops with the suggestions of the cultivation pattern.
Acknowledgment
This study has been conducted as an interior research project of Islamic Azad University- Isfahan (Khorasgan) branch No. 698. The author appreciates the university vice chancellor of research for its financial resources.