Analysis Of Tobacco Planting Soil Nutrient In Shaoyang County And Prediction Of Yield Per Mu Of Tobacco

As one of the important pillar industries in Shaoyang County,tobacco’s output and quality are closely related to the economic development of Shaoyang County.Taking Shaoyang County as an example,this paper analyzes the soil nutrient and texture characteristics of the tobacco-growing area in Shaoyang County.And then,the minimum redundancy maximum correlation analysis(m RMR)and the support vector machine recursive feature elimination method(SVM-RFE)are applied to the nutrient features sorting.After selecting nutrient features by principal component analysis(PCA),the optimal feature subsets were selected from soil nutrient indicators,which is applied to the support vector machine regression(SVR)and random forest regression(RFR)for tobacco yield prediction.By comparing and analyzing the prediction results of it,the tobacco yield analysis and prediction system are designed.The main study contents and findings are listed as follows:(1)Based on the analysis of soil texture and nutrient characteristics of 70 soil samples from tobacco-growing areas in Shaoyang county and soil grade and main nutrient content distribution and abundance of tobacco-growing areas,the correlation between soil nutrient indicators are analyzed by the Pearson correlation coefficient method.The results showed a significant correlation between soil nutrient indicators.(2)Applying the mRMR and SVM-RFE to sort the soil nutrient indexes of Shaoyang County and the PCA to select the soil nutrient index,the comparative analysis of the above feature selection methods showed that the PCA has better feature selection effects.(3)Optimal feature subsets obtained by principal component analysis is applied to the SVR and RFR of tobacco yield prediction in Shaoyang County,and the root mean square error(RMSE)was adopted for evaluation and comparative analysis.The results showed that the SVR prediction has a better predictive effect on tobacco yield per mu.(4)Based on principal component analysis,support vector machine and random forest,the system logic architecture,functional architecture,database and system interface for the tobacco yield analysis and forecasting system in the tobacco-growing area of Shaoyang County were designed,which has the functions of soil nutrient analysis and prediction,model training,tobacco yield forecasting,record management and user rights management.The results showed that there is a significant correlation between soil nutrients in the tobacco-growing areas of Shaoyang County.The PCA can effectively extract the characteristics of soil nutrient indicators,The SVR prediction and the RFR prediction have a better predictive effect for tobacco yield per mu,providing decision support for tobacco planting and management in Shaoyang County to a certain extent.