| Abstrakt: |
Investing in stock market is considered a high-risk monetary motion where investor may not take into version the factor of stock charge fluctuations or destroy their profits without gaining professional knowledge and investment experience. In addition, when working with stock information, people increase the importance of available and self-correcting information, a habit that runs counter to objective and reasonable investment decisions. In recent years, various methods of stock market forecasting have been proposed to enhance the quality and profitability of investor decision making. Recent machine learning models have been able to reduce the risk of stock market forecasting. However, diversity remains a major challenge in developing better learning models and in identifying intellectually valuable traits to further improve diagnosis. In this paper, we propose an optimal characteristic selection with hybrid machine learning classifier for stock market price forecasting (OFML-SMF) using social media and secondary data sources. First, we introduce a multi-objective earthworm optimization (MOEO) algorithm for optimal feature selection among multiple features i.e. called feature optimization which reduces the data dimensionality problem. Second, we develop a hybrid Quantile regression learning based deep belief neural network (HQR-DBNN) to classify the stock market price movement prediction which enhances the statistical metrics. Finally, to appraise the presentation of planned OFML-SMF model through the influential international stock market indices and the performance can be compare with the existing state-of-art models in conditions of accuracy, precision, recall and F-measure. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink