Prediction Of Grassland Forage Growth Based On Data Mining

As the country further accelerates the construction of ecological civilization,the grassland ecological problem has become a major issue of great concern to all sectors of society.Forage grass yield is one of the important indicators to determine the function of grassland ecosystem.For the forecasting of forage yield,how to improve the accuracy of forage yield forecasting has become the focus of current research.At present,based on data mining,a series of forage grass yield prediction models are proposed at home and abroad,but the accuracy of the model prediction results needs to be improved.Therefore,based on the data mining theory,this paper proposes a forage grass yield combination forecasting model,which can improve the precision of pasture yield.The research content of this paper is as follows:(1)Select a suitable prediction model.Based on the data mining theory,this paper mainly selects multiple linear regression analysis,principal component analysis and RBF neural network to establish a combined prediction model of forage and grass yield.In the process of model building,multiple regression analysis is used to predict the linear part,and principal component analysis is used to solve the possible collinearity problem.RBF neural network has the characteristics of simple structure,fast training speed and strong nonlinear approximation ability.To deal with the nonlinear part,multiple regression analysis can compensate for the information loss of RBF neural network prediction results and over-reliance on sample accuracy.Therefore,these two methods are selected for modeling.(2)Establish a forecast model for forage yield.In order to better predict the grass yield of pasture,based on the data mining theory and the advantages of the selected model,this paper proposes a forage grass yield combination forecasting model.The model uses a multivariate regression model to analyze raw data samples to obtain predicted values and residuals.The obtained residuals were fitted using a radial basis neural network,and the predicted data was corrected using the fitted residuals to obtain the final result.(3)Comparative analysis of multiple model prediction results.The temperature,precipitation,and sunshine historical data of the Ulan Buddhism grassland were selected to establish a model.SPSS software,multivariate regression model,principal component analysis and RBF neural network were used to complete the modeling.Compared with the forage grass yield forecasting model.The average relative errors were 18.17%,11.49%,6.946% and 4.398%,respectively.The results showed that based on the combination of multiple regression algorithm and RBF neural network,the prediction results were significantly improved in terms of fitting accuracy and prediction accuracy compared to the single prediction model.Therefore,the effectiveness of the combination forecast model of forage grass yield is was obtained,thus providing decision-making basis for grassland ecosystem protection.