Improved Grey Mixed Model And Its Application In The Prediction Of Network Public Opinion

In the current information age,the popularity and development of the internet has promoted the generation and dissemination of network public opinion.With the continuous increase of the number of internet users,the evolution of network public opinion has an increasing impact on social stability and development.Therefore,it is necessary to grasp the natural law of public opinion development through scientific data analysis,and make predictions and judgments on the evolutionary behavior of public opinion.In order to better carry out artificial intervention to guide or encourage the spread and development of public opinion.The construction of high precision network public opinion prediction model is a difficult problem.At present,the main modeling methods to solve this problem are combination model,hybrid model,integrated model and artificial intelligence model.Because the grey uncertainty of network public opinion data,the disturbance term of the inclusion has a great influence on the prediction accuracy of the model.In this paper,the gray buffer operator is introduced to preprocess the data to improve the prediction accuracy of the model.Due to the instantaneity and short duration of public opinion,network public opinion data has the characteristics of a small sample.In view of this feature,this paper uses a gray model to predict public opinion.However,the prediction ability of a single model is limited,and even after the improvement,the prediction accuracy is still not very high,but the hybrid model can solve this problem.Considering that the least squares support vector machine model in the artificial intelligence model is also suitable for small sample prediction and can better deal with nonlinear problems,the least squares support vector machine is used to modify the residual of the grey model in this paper,and finally establish an improved hybrid model of grey model and least squares support vector machine.It is verified that the hybrid model has higher prediction accuracy by MAPE and MSE.The specific work is as follows:(1)Because the change trend of public opinion data can't fully show the real change rule of data itself,the concept of buffer operator in grey theory is introduced,and the data sequence is preprocessed by using the geometric average weakening buffer operator(GAWBO),which makes the data sequence more smooth and regular,and is conducive to the simulation and prediction of the model,so as to improve the prediction accuracy.(2)In this paper,two hybrid models are established.First,an improved single gray model is established,namely the LGM(1,1)model and the TPGM(1,1)model optimized based on the initial value.Then,in order to solve the problem of poor prediction performance and low accuracy of a single model,the two models were corrected for the residual error using the least squares support vector regression machine.Finally,two hybrid models based on the improved gray model and the least squares support vector machine model were established,namely the GAWBO-LGM-LSSVM hybrid model and the GAWBO-TPGM-LSSVM hybrid model.Select a typical network public opinion data sequence(Baidu index of the popular keyword "IG wins the championship")to test the prediction effect of the model.According to the prediction accuracy evaluation index MAPE and MSE,the experimental results of each model are compared and analyzed step by step.It is proved that the hybrid model proposed in this paper has higher prediction accuracy than the classic single gray model and the improved single gray model.