Volume 30, Issue 2, 2021
DOI: 10.24205/03276716.2020.4101
Prediction of Financial Economic Time Series based on Group Intelligence Algorithm based on Machine Learning
Abstract
The prediction of financial economic time series can help investors avoid risks and obtain higher returns by forecasting the future price according to the historical transaction data of financial transaction varieties (such as stocks). However, the financial economic time series is a chaotic time series which is extremely complex, nonlinear, non-stationary and high noise related. Therefore, the prediction of financial economic time series is considered to be the most difficult topic in the study of modern time series. This paper mainly studies the prediction of financial economic time series based on Group intelligence algorithm based on machine learning, The advantages of this prediction can only improve the quality of clustering, reduce the calculation amount, improve the speed of operation, and make the regression analysis and prediction of financial time series more effective. This paper mainly uses machine learning method, support vector regression and SVR algorithm to study the financial economic time series prediction based on group intelligence algorithm based on machine learning. Support vector regression (SVR) is widely used in financial time series prediction, and it shows stronger prediction ability than traditional artificial neural network. This is mainly because SVR is a machine learning algorithm based on statistical learning theory, which has good nonlinear approximation, fast convergence, global optimal solution and strong generalization ability. The results show that the algorithm combines the prediction algorithm of multiple SVR models in five research samples, selects multiple SVR models in different training data sub group training, and uses reasonable weight combination to predict the results. The diversity of models is used to reduce the overall prediction error. The weights of each model are dynamically adjusted according to the latest prediction accuracy, so it has the adaptability and can deal with the problems caused by the non-stationary. The algorithm of mixing multiple SVR models can significantly improve the accuracy and generalization ability of financial time series prediction
Keywords
Machine Learning, Group Intelligence Algorithm, Financial Economic Time Series Prediction, Svr Algorithm