Analysis of Feature Importance Based on Random Forest for Stock Selection

Yiru Wang

2024

Abstract

As a matter of fact, feature importances are quite crucial for stock selection. This study investigates the application of feature importance analysis using a random forest model for stock price prediction, focusing on the semiconductor and public utility sectors. Given the inherent volatility and complexity of stock markets, traditional linear models often fall short in capturing non-linear relationships and interactions among variables. This research leverages machine learning techniques, particularly random forests, to identify key technical indicators that significantly influence stock performance. By analysing data from Yahoo Finance, the study incorporates a diverse set of factors, including momentum and volatility indicators, to assess their predictive power. The findings reveal that volatile sectors like semiconductors benefit from indicators such as short-term trends and volatility, while stable sectors like public utilities are more influenced by sector-specific conditions. These results provide a comprehensive framework for factor selection in stock analysis, highlighting the importance of a tailored approach based on sector characteristics. Future research should expand this analysis to include macroeconomic conditions and sentiment indicators, offering a more holistic view of the factors driving stock prices and enhancing investment strategies.

Paper Citation

in Harvard Style

Wang Y. (2024). Analysis of Feature Importance Based on Random Forest for Stock Selection. In Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI; ISBN 978-989-758-726-9, SciTePress, pages 374-379. DOI: 10.5220/0013233500004568

in Bibtex Style

@conference{ecai24,
author={Yiru Wang},
title={Analysis of Feature Importance Based on Random Forest for Stock Selection},
booktitle={Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI},
year={2024},
pages={374-379},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013233500004568},
isbn={978-989-758-726-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI
TI - Analysis of Feature Importance Based on Random Forest for Stock Selection
SN - 978-989-758-726-9
AU - Wang Y.
PY - 2024
SP - 374
EP - 379
DO - 10.5220/0013233500004568
PB - SciTePress