Stock price prediction based on Weighted Meta-Extreme Learning Machine
Yanbing Song, Shaofei Zang, Jianwei Ma, Huimin Li, Jinfeng Lv · 2024
With the continuous development of China's stock market, stock forecasting has gradually become a focus of attention. Aiming at the Meta-ELM algorithm in the field of stock price trend prediction ignores the problem of the existence of the time characteristics in the time series data and the difference in the effect of the base ELM, this paper proposes a short-term stock price trend prediction model based on the Weighted Meta-Extreme Learning Machine (WMeta-ELM). The method first improves the Meta-ELM model by making the number of time steps of the stock data equal to the number of base ELM models, enabling the model to extract time series features from the stock series. In order to better capture the dynamics of the market, the method also introduces a weighting approach that assigns different weights to each base-ELM model to guide the model to learn different information at different time points. In order to validate the effectiveness of the proposed model, we compare it with five traditional and advanced models in real stock trading data. The results show that the proposed WMeta-ELM model is more feasible and effective than Meta-ELM.