| With the iteration of artificial intelligence technology and the refinement and clarity of data in the financial fields,quantitative investment and precise investment in stocks have become more and more realistic.The purpose of this project is to design a stock prediction and stock recommendation system based on machine learning models,make predictions on stock trends based on daily stock data and stockholder's sentiment,and give certain recommendations for stockholders' investment,and system can provide users dominant stocks based on prediction results and similar to their optional stocks.This system combines stock technical indicators and stockholders' sentiment dual impact factors to make predictions on stock price.The brief steps are first analyse stockholder's sentiment based on deep learning models,and then combine with basic technical indicators to feed into the prediction model to complete stock price prediction.This system includes account information management module,optional stock management module,data collection and pre-processing module,stock prediction module,stock real-time market module and stock recommendation module.My work is as follows: Collection of historical data required by the project,mainly including historical stock data,historical stockholder's sentiment data,labeled public sentiment text and Chinese corpus.Model architecture and training,mainly including training the LDA topic model and implementing stock classification,and then conducting stock personalized recommendation testing and achieving good results;transforming and training the BERT-WWM model,and then testing the model on the labeled Chinese dataset and achieving good results;constructing and training the predict model,and then testing the model on the real data set and achieving good results.System design and implementation mainly includes system requirements analysis,system design,system implementation and system testing.This project uses Django,Nginx,u WSGI,etc.for development and deployment,and the final system meets the expected requirements.However,there are still parts that can be further optimized in the project,and I hope that they can be continuously improved in future work. |
