Research On The Prediction Of Digital Currency Prices Based On Data Mining Social Media Data

In the natural sciences and social production operations,a large number of decisions are involved.Decision-making helps us to choose the most favorable decision under a large number of uncertain situations.Based on the Stacking integrated learning framework,this paper studies the impact of social media data on the digital currency market.The main work content includes the following points:(1)This article first conducts an in-depth study on the development of digital currency and the influencing factors of data currency prices.Secondly,it explains the commonly used methods of social media data mining,and summarizes previous scholars' research on social media data.(2)This paper systematically sorts out investment theory,feature selection technology and data mining technology.(3)This article uses web crawler technology to obtain the research sample data of this article through the Python self-encoding method.(4)For the social media data in this article,this article quantifies 19 social mediarelated feature variables based on the characteristics of each platform data.At the same time,this article analyzes the importance of feature variables through the feature importance ranking method.(5)Through experimental comparison,this paper found that in the data set of this article,the prediction accuracy of the logistic regression algorithm(61.24%)is the lowest.In the previous data mining model,the prediction accuracy of the XGBoost algorithm is higher,reaching 75.07%,which is compared with LR.In terms of algorithm,the accuracy rate has increased by nearly 14%,and the prediction accuracy rate of the integrated learning framework in this article is 1.32% higher than that of XGBoost...