Application Research Of Multi-Domain Based On MF-DCCA

Modern science has become a data-intensive discipline.The demand for innovative solutions for data management,data sharing and discovery of novel data sets is increasing.Because the Multi-Fractal De-trended Fluctuation Cross-Correlation Analysis(MF-DCCA)method can well characterize the nonlinear fluctuation correlation between two non-stationary time series.This article makes a non-linear correlation study in the fields of drug development,industrial workers,and stock markets.In the field of drug development,this article is the first to study the non-linear crosscorrelation between drug molecular properties and aqueous solubility using pseudo time series.The experimental results show that there is a weak long-range cross-correlation between the drug molecular weight and aqueous solubility of drug molecules;there is a weak power-law cross-correlation between the drug molecular properties of other drug molecules and aqueous solubility;the width of sliding window versus time-varying Hurst The robustness of the index is strong.This work can provide a new feature selection method for the multi-scale feature input task of deep learning drug molecular property prediction.In the field of industrial workers,this article investigates the research literature and network search of industrial workers through the combination of network big data and MF-DCCA method.A statistical analysis of the literature fund funding,literature publication year,and major research institutions related to the industrial worker team conducted.This article proposes a framework for analyzing industrial workers' teams through the integration of natural language processing,Baidu index and MF-DCCA.Keywords extracted from CNKI literature,and keywords used to search Baidu index time series.MF-DCCA method used to analyze the correlation between keyword-based time series and industrial worker time series.This work can help the government understand public opinion,study the status quo and take relevant measures to respond to public opinion.In the field of stock market,this article proposes a forecasting framework for stock price trends.It includes data collection,correlation analysis,and prediction of stock price trends.The experimental results show that there is a long-range cross-correlation between the big data on the Internet and the trend of stock price fluctuations;the Baidu index before and after the addition of the Baidu Index shows that it does not contribute uniformly to the performance of stock price fluctuations.The possible reason is that investors' irrational behavior makes the Baidu index mismatch with the actual stock market information or the Baidu index's integration of public search behavior data is too poor.This work can help niche investors quickly determine the trend of stock prices.