Research on the Prediction of the Development of ESI Disciplines in Universities Based on ARIMA and GM(1,1)Models
[Objective/Significance]Subject construction is a key aspect for universities to enhance the quality of education and plays an important role in supporting scientific research.This article adopts mathematical statistical modeling to explore a scientifically effective method for predicting the time it takes for a potential subject to enter the top 1%in ESI rankings.This has significant guidance implications for institutional subject development planning.[Methods/Processes]Based on the ESI database,this paper obtains the citation frequency and ESI shortlisting threshold of the four potential disciplines of the target institution,establishes the time series,and creates a prediction model:first introduce the conversion coefficient to remove the differences between different databases and make them comparable,and then fit the GM(1,1)model and ARIMA model respectively to predict the citation frequency and ESI shortlisting threshold of the target academic institution,and find the time when the citation frequency of the subject of the target institution catches up with the ESI shortlisting threshold,that is,the predicted shortlisting time.By using mean absolute percentage error(MAPE),mean absolute error(MAE)and root mean square error(RMSE)to evaluate and compare the fitting and prediction effect of the model,the model fitting and prediction effect were evaluated according to the three indicators of MAPE,MAE and RMSE,so as to provide a reference basis for the discipline construction and long-term development planning of the school.[Limitations]The limitation of the study is data from only four disciplines in the target institution.Additional data from other institutions and more disciplines are needed to validate the predictive performance of the model.[Results/Conclusions]The fitting effect and prediction effect of ARIMA model are better than those of GM(1,1)model.The biology and biochemistry disciplines of the target institution will be in the top 1%of ESI in the coming months;Immunology has the potential to be shortlisted in the top 1%of ESI,but the shortlisting time may be slightly delayed;The disciplines of molecular biology and genetics and neuroscience and behavior are still far from being shortlisted.
ESIIncitesPotential DisciplineGray ModelARIMA Model
万方数据
维普
