融合情感分析与 SVM_LSTM 模型的股票指数预测

由于股票市场变化存在着多因素、非线性、时变性等特点，传统预测模型忽视了股指波动影响因素特征提取的合理性与准确性，导致预测效果不理想。鉴于此，提出了融合情感分析和 SVM_LSTM 特征提取模型的股指预测方法以提高股指预测精度，将 SVM 和 LSTM 方法相结合建立 SVM_LSTM 模型，提取影响股指波动的情感极性特征、涨跌趋势特征以及股票技术指标特征，进而弥补影响股指波动的存在因素实现股指预测。通过与传统股指预测方法相比较，该方法实验结果的 MSE（均方差）达到了 0.172 2，比传统模型的均方差缩小了约0.083 7，证明了该预测方法在准确度上效果更好。

Stock Index Prediction Based on SVM _LSTM Model with Emotion Analysis

Due to the multi-factor，non-linear and time-varying characteristics of stock market changes，traditional prediction models ignore the rationality and accuracy of extracting the characteristics of factors that affect stock index fluctuations，resulting in unsatisfactory prediction effect. Therefore，this article puts forward the integration analysis and SVM_LSTM emotional feature extraction model of stock index prediction method to improve the predictive accuracy of the stock index. By combining the SVM and the LSTM method SVM_LSTM model is established to extract the influence of stock index volatility emotion polarity features，price trend and the characteristics of stock technical indicators so as to make up for the factors affecting the existence of the stock index fluctuation index prediction.Compared with the traditional stock index prediction method，the MSE（mean square deviation）of the experimental results of the proposed method reached 0.172 2，which is about 0.083 7 smaller than the mean square deviation of the traditional model，proving that the proposed prediction method has better accuracy effect.