基于XGBoost 农业经济产业结构分析———以广东省中山市为例.

By constructing the XGBoost model, this paper takes the output of planting, fishery, animal husbandry and forestry in Zhongshan over the years as the independent variable and the total output of agricultural economy as the dependent variable for regression analysis. The research shows that the average growth rate of planting industry output value over the years is 3. 45%, the average growth rate of forestry output value is 2. 16%, the average growth rate of animal husbandry output value is 0. 46%, and the average growth rate of fishery output value is 9. 69%. The model fitting curve shows that all agricultural industries have promoted the positive growth of Zhongshan's total agricultural economy. From the analysis of the characteristic correlation parameters of the XGBoost model, it is concluded that the fishery industry has the largest contribution to the total agricultural output value of Zhongshan City, accounting for about 59. 8%, the planting industry accounts for about 35. 5%, the animal husbandry industry accounts for 4. 2%, and the forestry industry accounts for 0. 5%. It shows that fishery and planting industry are the pillar industries of agricultural economy in Zhongshan City. Through comparative analysis with other models, the evaluation indicators proposed in this paper are 0. 715 5, 0. 999 9, 6. 499 1×10-6 through root mean square error, R², and mean Poisson deviation regression Loss, respectively. This method is superior to other models, which verifies that the proposed XGBoost model has high fitting accuracy and robustness for the fitting and feature correlation analysis of the output value data of various industries in Zhongshan over the years. This method can provide a good regression fitting analysis model for the agricultural industry. [ABSTRACT FROM AUTHOR]