Comparison Of Different Time Series Analysis Methods In Predicting The Number Of Public Library Institutions In China

Public cultural service has become an important part of socialist culture with Chinese characteristics.The public library is an important carrier of public cultural services,which helps to spread scientific and cultural knowledge,enhance the overall knowledge of citizens,improve the comprehensive quality of the whole people,promote the development of modern education,and create a good cultural atmosphere,so as to respond to the Third Plenary Session of the 18 th CPC Central Committee clearly put forward the construction of modern public cultural service system and the19 th National Congress of the Communist Party of China put forward the cultural rejuvenation strategy.The number of public library institutions is the most intuitive embodiment of the development of public libraries.Through the prediction of the number of public library institutions in China,we can understand the importance and development of public cultural service ability in China.In order to grasp the dynamics of the number of public libraries in China as a whole,understand the current development level and future development trend of public libraries in China,make public libraries play a better role in public cultural services,effectively promote the rapid development of public cultural services in China,and have important guiding significance for the prediction and analysis of the number of public libraries in China.This paper selects the historical data of the number of public library institutions in China from 2001 to 2020 as the research data,takes the data from 2001 to 2017 as the training sample,and takes the data from 2018 to 2020 as the test sample to predict the number of public library institutions in China from 2021 to 2023.Firstly,the trend extrapolation model,exponential smoothing model and grey model are used to make the prediction.Secondly,on the basis of the above single model,four different weight determination methods are used to build the equal weight method combination model,the simple weight method combination model,the variance reciprocal method combination model and the mean square reciprocal method combination model to do the prediction.Finally,the sum of square error,average absolute percentage error,Theil inequality coefficient and prediction accuracy are used to compare the prediction results of the above models.Through the empirical prediction analysis,it is concluded that the combination model can improve the prediction effect of a single model to a certain extent,the prediction effect of combination model constructed by different weights will be different,and the selection of model evaluation index will also affect the evaluation of model prediction results.In the prediction analysis of the number of public libraries in China,the exponential smoothing model is the most unsatisfactory,the trend extrapolation model is the best model in the single model,and the simple weighted combination model is the best model in the combination model.