Forecast On China-ASEAN Exchange Rate Based On Multiple Forecast Model

Exchange rate is a very important regulating lever in international trade.Whether it can accurately predict the long-term trend of exchange rate has important reference value for the venture capital and trade between China and ASEAN countries.The analysis and prediction of the long-term trend of China-ASEAN exchange rate is also the main direction of the financial sector in the financial sector of the China-ASEAN Marine big data platform.According to the existing literature,there are two main methods to predict the exchange rate:basic analysis and technical prediction.This paper selects Singapore as the research object of ASEAN countries and USES these two methods to predict the long-term and short-term official exchange rates of Singapore.When using the technical analysis method to predict the long-term exchange rate,the impact factor data associated with the annual exchange rate are obtained by using PCA at first,and the data preparation process is completed.Next,the prediction model was established.According to the small sample characteristics of annual exchange rate data,the support vector machine regression algorithm was used to predict exchange rate.In order to solve the problem of slow parameter convergence,the FOA-SVR exchange rate prediction model based on drosophila optimization algorithm was proposed.In order to avoid the limitation of single prediction model,BP neural network model is introduced in this chapter,and combined with FOA-SVR exchange rate prediction model in parallel,FOA-SVR-BPNN exchange rate prediction model is proposed.The experimental results show that the combined model is better than the ordinary SVM regression model.When using the basic analysis method to predict the short-term trend of the exchange rate,the ARIMA model based on time series is used.The short-term historical monthly data is taken as the time series.After eliminating the trend and seasonal factors,the series is regressive predicted.Based on the ARIMA model,this paper takes the predicted ARIMA value as the linear predicted value,and the residual of the predicted ARIMA value as the input of the BPNN model,makes nonlinear prediction for the residual,and finally combines the linear predicted value with the nonlinear prediction result,and proposes the ARIMA-BPNN prediction model.The prediction effect of the improved ARIMA model is better than that of ordinary ARIMA model.Finally,the research results of this paper are applied to the financial module of China-ASEAN ocean big data platform,respectively realizing the prediction function of long-term exchange rate and short-term exchange rate,and providing users with prediction reference and data services.