Evaluation And Prediction Of Agricultural Production Potential In Heilongjiang Province

Agriculture is the country’s primary industry,supporting the country’s social activities and scientific and technological development.Ensuring the effective supply of agricultural products has always been the top priority of the country’s "Three Rural" work.Heilongjiang is the main province of agricultural production in China.It is of great significance to evaluate and forecast the agricultural production potential of Heilongjiang Province for ensuring national food security,optimizing agricultural production layout and promoting sustainable development of resources.This paper comprehensively analyzes the current status of agricultural production and development in Heilongjiang Province,constructs the evaluation system of agricultural production potential,uses the combination weight method to construct the evaluation model to evaluate the agricultural production potential of Heilongjiang Province,introduces the graph neural network algorithm TGCN to predict the agricultural production potential index of Heilongjiang Province,and puts forward targeted suggestions for the evaluation and prediction results.The research results are as follows:(1)The evaluation system and evaluation model of agricultural production potential are constructed.Considering the status of agricultural production and relevant factors affecting agricultural production,an evaluation system of agricultural production potential composed of 36 indicators is proposed from the 5 dimensions included grometeorological potential,land resource potential,agricultural output potential,agricultural development potential and science and technology input potential;The evaluation system and evaluation model of regional agricultural production potential are formed by using the optimal combination weight method based on the sum of squared deviations and synthesizing the weights of the single weight method.(2)The entropy weight method is improved and the random forest weight method is proposed.This paper analyzes the shortcomings of the standard entropy weight method in the time series system,and uses KL divergence to replace the entropy to improve the entropy weight method,so that the improved entropy weight method comprehensively considers the specificity of the index distribution in the system when weighting the index,and ensures the scientificity and systematicness of the weighting result;The factor analysis method in the field of statistics and the random forest algorithm in the field of machine learning are integrated.According to the feature importance obtained by the random forest algorithm in training,the random forest weight method is established to ensure the interpretability of the weighting results.(3)The agricultural production potential of Heilongjiang Province is evaluated.According to the agricultural production data of Heilongjiang Province,the agricultural production potential of Heilongjiang Province from 2000 to 2019 is comprehensively evaluated,and K-means algorithm is used for clustering of agricultural production potential.The results show that the potential index of agricultural production in Heilongjiang Province shows a fluctuating upward trend,which is in a rapid upward range from 2000 to 2004,a downward range from 2004 to 2006,a rapid upward range from 2006 to 2012 and a slow upward range from 2012 to 2019.(4)The potential level of agricultural production is predicted.DPSIR is used to analyze the causal chain of agricultural production potential evaluation indicators,and the adjacency matrix between features is obtained.The deep learning algorithm T-GCN is used to predict the agricultural production potential of Heilongjiang Province from 2020 to 2024,and the obstacle degree of each index in this interval is calculated.The results show that T-GCN performs well in the evaluation task of agricultural production potential,and the agricultural production potential of Heilongjiang Province will continue to rise.The cultivated land area of agricultural reclamation system,the proportion of financial investment in agriculture,forestry and water affairs and the proportion of agricultural employed population are the main obstacle indicators of agricultural production potential of Heilongjiang Province from 2020 to 2024.According to the prediction results,targeted suggestions are given to improve the agricultural production potential of Heilongjiang Province from three aspects: controlling the loss of agricultural population,ensuring the safety of agricultural land and promoting the high-quality development of agriculture.