Predicting Chinese Stocks Using XGBoost-LSTMAttention Model
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Forecasting is a popular topic in the stock market. In recent years, many scholars have utilized machine- anddeep-learning models in this field. However, many stock forecasting models suffer from problems ofinformation overlap in stock trading data and a relatively simple structure of the prediction model. To overcomethese issues, we built a stock forecasting model based on extreme gradient boosting (XGBoost), long shorttermmemory (LSTM), and attention (XGBoost-LSTM-Attention). XGBoost is used to extract importantinformation from stock data, and the LSTM combined with the attention mechanism can enhance stockprediction performance. To verify the feasibility and effectiveness of XGBoost-LSTM-Attention, we selected14 Chinese stocks from different industries for the prediction experiments and compared their performancewith those of existing models. The experimental results showed that the average root-mean-square error valueof the XGBoost-LSTM-Attention model for the different stock datasets was the smallest (0.012); the averageR2 value (0.96) and average accuracy (66.1%) were the highest.
Abstract
Forecasting is a popular topic in the stock market. In recent years, many scholars have utilized machine- anddeep-learning models in this field. However, many stock forecasting models suffer from problems ofinformation overlap in stock trading data and a relatively simple structure of the prediction model. To overcomethese issues, we built a stock forecasting model based on extreme gradient boosting (XGBoost), long shorttermmemory (LSTM), and attention (XGBoost-LSTM-Attention). XGBoost is used to extract importantinformation from stock data, and the LSTM combined with the attention mechanism can enhance stockprediction performance. To verify the feasibility and effectiveness of XGBoost-LSTM-Attention, we selected14 Chinese stocks from different industries for the prediction experiments and compared their performancewith those of existing models. The experimental results showed that the average root-mean-square error valueof the XGBoost-LSTM-Attention model for the different stock datasets was the smallest (0.012); the averageR2 value (0.96) and average accuracy (66.1%) were the highest.