ARIMA-BP Neural Network Combination Model for Prediction of Bovine Brucellosis Incidence Rate in Xinjiang
Xinjiang is the second largest pastoral area in China with high livestock rearing,and also a traditional endemic area for brucellosis.Data on the incidence rate of bovine brucellosis from January 2017 to February 2023 were selected as the training sam-ple set.The ARIMA model and BP neural network method were used to model the training sample set,respectively,and their fitting effects were evaluated and analyzed.The results showed that the time series ARIMA model was suitable for linear prediction,but the fitting effect was poor in case of large fluctuations in the series and in the presence of outliers.The BP neural network demonstrated strong persuasive power in fitting nonlinear fluctuation patterns and had a clear advantage in handling outliers.Therefore,drawing on the advantages of the ARIMA model and ARIMA-BP neural network in fitting the incidence rate of bovine brucellosis in Xinjiang,the combined ARIMA-BP neural network model was constructed and fitted to the training sample set.Furthermore,the above three models were applied to predict the incidence rate of bovine brucellosis in Xinjiang's breeding sector from March 2023 to February 2024,and the results were compared with the actual values.The results demonstrated that,the combined ARIMA-BP neural net-work model was better than the two single models in both the fitting of the training sample set and the prediction of the future inci-dence rate,and was feasible in predicting the future development trend of bovine brucellosis in Xinjiang.
万方数据
维普
