Arif Mudi Priyatno, Triyanna Widiyaningtyas, Didik Dwi Prasetya, Wahyu Caesarendra
Stock price prediction is a challenge in financial analysis due to the complex and volatile nature of market data. Conventional prediction models often experience a decrease in accuracy due to high feature dimensions, so they require efficient feature selection methods. Recursive Feature Elimination, one of the feature selection methods that can be used, has limitations in handling relevance and redundancy at the same time, and has not considered the relevance of features to the prediction target by integrating information from more than one model. To address this, this study proposes an integration between the Minimum Redundancy Maximum Relevance based Relevance-redundancy Difference (mRMR) and the Coefficient-Average Based Recursive Feature Elimination modification (CA-RFE). The proposed method aims to optimize feature selection by using mRMR to filter out redundant features, as well as CA-RFE to rank features based on the average normalized coefficients of Support Vector Regression and Linear Regression. The research was conducted using stock price data from yahoo finance for 14 years. Features include historical data and technical indicators. The evaluation results using the Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) metrics show that the proposed method provides a significant reduction in prediction error compared to previous methods. In PT Bank Central Asia Tbk (BBCA), the MAPE value is 1.11%, lower than other methods. The proposed integration of mRMR and CA-RFE has succeeded in improving the performance of machine learning models in predicting stock prices. © This article is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License. License details: https://creativecommons.org/licenses/by-sa/4.0/
Department of Electrical Engineering and Informatics, Universitas Negeri Malang, Malang, Indonesia; Department of Mechanical and Mechatronics Engineering, Faculty of Engineering and Science, Curtin University Malaysia, Sarawak, Malaysia
