Study On Combination Forecast Of The Number Of Domestic Tourists In China Based On Singular Spectrum Analysis

Tourism is a word that people are not unfamiliar with.Its vitality is vividly exhibited in our lives.Nowadays,with the continuous development of social economy,people's living standards have been greatly improved.After meeting the external material needs,they have gradually increased their pursuit of the inner spirit.Today,the global tourism industry has entered a golden age of rapid development.China is the world's second largest economy and the largest developing country,with the largest population in the world,its national consumption behavior is very active.Since China's economic development began to enter the "new normal”stage in 2012,the contribution rate of consumption activities to economic growth has increased significantly,which is also an important manifestation of economic development.At the same time,in the development of the national economy in recent years,tourism has gradually become an indispensable and important industry.Tourism development has become a big market,and the number of tourists continues to grow.In 2018,the added value of China's tourism industry accounted for more than 11%of GDP.Against such a background,in order to ensure the comprehensive and robust sustainable development of the domestic tourism industry,to ensure the continuous improvement of the quality of life of the people,and to promote the rapid development of the national economy,scientifically and accurately predicting the number of tourists in advance will be an important part of planning tourism development,and finding a reasonable and effective forecasting model is the key to ensuring that the forecasting results are realistic.In this paper,we use the statistical data of domestic tourists in the Chinese Statistical Yearbook 2019,introduce singular spectrum analysis technology to remove noise from the original time series,use ARIMA model,BP neural network and Holt-Winters method to model and predict the number of tourists based on noise reduction sequences.Then based on the prediction results of the single model,construct two combined prediction models and comprehensively analyze the effects of data preprocessing and model combination on the prediction result of number of tourists.The results indicated that,first,the singular spectrum analysis technology has a significant effect in reducing sequence noise and extracting the main information of the sequence,the prediction accuracy of the model has improved after preprocessing,which is reflected in that compared with the original tourist time series,the prediction based on the reconstructed sequence processed by singular spectrum analysis can produce smaller prediction errors.Second,compared with the single model,the prediction performance of the combination model is significantly improved.The combination model constructed in this paper has extremely high prediction accuracy and stable prediction performance.The weighted average combination method has the best effect,and the relative error on the test set is only 0.66%.Apply this combination prediction model to forecast and analyze the number of domestic tourists in China is feasible and effective.The results of this study prove the value of singular spectrum analysis and combination forecasting in the prediction of number of tourists,provide model support for the prediction and analysis of the national and regional tourist numbers and have guiding significance for scientific prediction of regional tourism development.Compared with the existing research,the contributions of this paper are mainly reflected in the following two aspects.In data processing,singular spectrum analysis is used to reduce noise on the original sequence of tourist numbers,extract the main characteristics of the data,reduce the irregularity of the original data,and have a significant effect on improving the prediction performance of the model.In terms of model selection,the shortcomings of the model simplification of some researches are broken.In this paper,three models that have been proven to work well in previous studies(ARIMA,BPNN,Holt-Winters)are selected as benchmarks,and two combined prediction models are constructed based on the single model.The scope of model selection for the prediction of number of tourists is effectively expanded,which can effectively improve the accuracy and stability of prediction.