Stock Price Forecast Based On SSA-MOGOA Algorithm

Time series prediction models have been widely used in recent years and have begun to play an important role in various fields,providing effective reference information for all aspects of the future development of the industry.In the stock market,most of the stock price time series information can be inferred and predicted by using the forecasting model established by statistical methods,which can avoid the complicated data acquisition process and a large amount of data.The interference of redundant information has been favored by most scholars.At the same time,stock price prediction is also of great significance to investors investment tendencies and strategies.The accuracy of the prediction model will directly affect investors'expectations of stock investment returns and risk assessment.However,the traditional economic forecasting models do not take into account the limitations of the individual forecasting models and the limitations of the data preprocessing used,resulting in the poor accuracy of these individual models in economic forecasting,so a new accuracy A high,easy-to-implement stock price prediction model is of great reference value to investors.In order to overcome these shortcomings of individual model predictions,on the basis of summarizing the existing research at home and abroad,a combination model combining data preprocessing technology,combined prediction and multi-objective optimization algorithms was developed to use stock historical price time series data.As a basis,one-,two-,and three-step forecasts were made for the future price of the stock.Because time series data in the financial field is usually very complex and special,and generally exhibits randomness,non-linearity,etc.,this article first pre-processes the stock historical price data,using EMD,CEEMD,EEMD,and SSA respectively.Based on the error comparison of the prediction results,SSA was finally selected as the pre-processing method of the combined forecasting model in this paper.Secondly,four common models of stock forecasting,ARIMA,BPNN,ENN,and ELM,were selected as separate models of the combined model.Finally,MOGOA multi-target the locust optimization algorithm determines the weight parameters of each individual model in the combined model.The data selected for this empirical study are from three representative educational institutions with long-term existence in the stock market:New Oriental Education Technology Group,Zhengbao Distance Education Group and Good Future Education Group.The reason why the stock data of the education industry is selected as the tested data to estimate the effectiveness of our proposed model is because the education industry has gradually become a popular investment industry due to the continuous development of the country and the strong support of the country in recent years,which is favored by investors.The empirical research mainly draws the following conclusions:? Based on multiple comparisons and analysis,the performance of the new combination model is significantly better than other benchmark models.By comparing the results of the prediction validity test,we find that our model performs best among all models applied in the experiment,has excellent prediction performance,can generate stable investment retun1s,and leads to a significant reduction in investment risk;? after The denoised processed separate model has improved the accuracy of stock prediction experiments to different degrees,and the singular spectrum analysis technology based on decomposition and reconstruction is used to obtain the basic characteristics of the time series by removing high-frequency signals.The contribution of the accuracy of the prediction results is higher than that of several other data preprocessing methods,whether in one-step,two-step or three-step prediction;? The multi-objective locust optimization algorithm optimizes the weight coefficients of each model in the combined model to obtain A combined model of high accuracy and high stability is presented.It is certain that the extensive application of this model will help investors to manage funds and make reasonable investment decisions.The innovations of the article mainly include the following two aspects:First,the application of decomposition and reconstruction strategies and the use of data pre-processing methods,by removing the main features of the original data extracted by high-frequency signals,make the prediction more accurate.Decomposing the original power data and reconstructing it into a filtered sequenee can eliminate the irregularity and uncertainty of the data and achieve better prediction performance.Second,the multi-objective locust optimization algorithm is applied to give the individual models in the combined model the best Excellent weighting factor.The shortcomings of the article are mainly reflected in two aspects:First,the data for the model test in this article are from three listed companies in the education industry.Versatility and portability are yet to be tested.Other scholars ean use stock prices in other industries.The data proposes improvements to the model.Second,this Paper ouly analyzes the time series of historical stock price data and does not perform in-depth analysis.It is recommended that other scholars add economic indicators and other factors that affect stock prices to conduct correlation analysis.