Wheat Yield Forecasting:A Machine Learning Approach Based On Meteorological Factors

In recent years,with the continuous improvement of machine learning technology,the forecast of wheat production based on meteorological data has gradually developed.In view of the limitations of traditional methods for predicting wheat yield,the use of machine learning prediction core algorithm to improve the prediction accuracy of wheat yield will once again promote the progress of meteorological prediction.The machine learning regression algorithm is usually used to deal with the "function approximation"problem,and the key impact factors are selected by feature extraction to construct the prediction model.This paper studies the meteorological prediction techniques of wheat traits and yield based on machine learning regression algorithm.According to the agricultural data of various traits and yields of wheat growth period in accordance with the national district test standards of Zhoukou City Academy of Agricultural Sciences,and the corresponding meteorological data collected based on ground meteorological observation points,the total collection of parts from 2007 to 2017 was collected for nearly 10 years.The data content,through data preprocessing and significance test,studied the wheat yield meteorological prediction model based on machine learning algorithm,and realized the meteorological comprehensive analysis of wheat.The article uses Pearson correlation analysis to screen key impact factors,including correlation analysis of key trait factors affecting wheat yield,correlation analysis of key meteorological factors affecting wheat traits,and correlation analysis of key meteorological factors affecting wheat yield.Based on the machine learning regression algorithm,a prediction model based on K-nearest neighbor,support vector machine regression,linear regression,decision tree,random forest and gradient lifting tree is established to realize the meteorological prediction of wheat yield and its traits.After the adjustment,the prediction effects of different models are analyzed,and the performance indicators such as mean square error and root mean square error are evaluated.The results show that the linear regression model is better,and the root mean square error of the prediction model is only 0.027058.Through the learning curve to fit the model,the linear model has the problem of"over-fitting".Two linear regression models of ridge regression and lasso regression are introduced,and the hyperparameter traversal is performed.The alpha is 0.00009.The Lasso model was used to predict the wheat yield and the root mean square error of the model was only 0.000159.It is indicated that the study of meteorological prediction of wheat yield is feasible using machine learning algorithms,and can effectively improve the accuracy of prediction and reduce errors.