Abstract:Abstract: With the rapid development of mobile internet technology, how to use public stock opinion to predict the closing price timely and accurately has important guiding significance for investors and business managers. This paper proposes an IWOA-Elman predicted model based on CEEMDAN algorithm. Text mining technology is used to mine and quantify stock opinions of SSE 180 Index. Boruta algorithm is used to select the important attributes to reduce the complexity of attribute set. CEEMDAN algorithm is used to realize the decomposition of the attribute sequence by adding a certain number of specific variances of white noise to the original closing price sequence. At the same time, the initial weights and thresholds of Elman neural network are optimized by
enclosing prey, spiral updating position and searching prey of WOA algorithm in the process of iteration, and the adaptive weights are used to improve WOA algorithm. The experimental results show that, compared with SVR, BPNN, Elman and GA optimization algorithms, the combined prediction model IWOA-Elman based on CEMDAN data set not only overcomes the slow convergence speed of WOA algorithm, but also effectively combines the advantages of infinite approximation of Elman neural network. Its average absolute error (MAPE) is 1.44531%, MAPE is reduced 3.49789% compared with the single Elman neural network. This study provides an effective experimental method for forecasting the public opinion of stock market network.