The Impact of Machine Learning Algorithms on Improving the Predictive Ability of Accounting Information as Business Partners: An Empirical.Evidence from Egyptian listed firms

This study is designed to provide empirical evidence on the impact of machine learning algorithms on the Predictive Ability of Accounting Information, by investigating the predictive ability of machine learning algorithms as opposed to traditional prediction models that rely on accounting information, on the accuracy of stock price predictions. It also aims to investigate whether machine learning algorithms' predictive ability outperforms traditional prediction models' predictive ability in cash-holding prediction models. Additionally, this study seeks to explore the practical potential of integrating machine learning algorithms and accounting information into prediction models to improve the predictive ability of accounting information. To fulfill this study's expected objectives, the researcher used several approaches.  Using a case study approach compared event study approach, the comparative analysis results revealed that accounting information's predictive ability was more accurate than the machine learning algorithms’ predictive ability in stock price predictions. This suggests that using machine learning algorithms does not necessarily result in better prediction performance and that machine learning algorithms are not the replacement for accounting information in financial predictions.
 
using the empirical approach, 564 firm-year observations from 2019 to 2022 are analyzed to predict cash-holding. The researcher employed several algorithms such as decision trees, support vectors, and K-nearest neighbor, compared to multiple linear regression based on accounting information as prediction models. The empirical results showed that decision trees as complex algorithms were proven to yield higher accuracy. While, other prediction models, MLR, KN, and SV had (RMSE) and low R², which indicates the Low accuracy of predictions. Also, the empirical results reported that accounting information significantly affects the accuracy of    algorithms regarding cash-holding . Hence, the current study adds empirical evidence to related previous  review through emphasizing the complementary nature between the roles of machine learning algorithms and accounting information in financial predictions. Machine learning algorithms must be considered a supporting tool to enhance the predictive ability of accounting information, not a replacement for it. At the same time, accounting information improves the accuracy of machine learning algorithms' predictions. Therefore, it has become necessary for accountants to master  skills to maintain their jobs and assume new roles in the  of artificial .
This study is designed to provide empirical evidence on the impact of machine learning algorithms on the Predictive Ability of Accounting Information, by investigating the predictive ability of machine learning algorithms as opposed to traditional prediction models that rely on accounting information, on the accuracy of stock price predictions. It also aims to investigate whether machine learning algorithms' predictive ability outperforms traditional prediction models' predictive ability in cash-holding prediction models. Additionally, this study seeks to explore the practical potential of integrating machine learning algorithms and accounting information into prediction models to improve the predictive ability of accounting information. To fulfill this study's expected objectives, the researcher used several approaches.  Using a case study approach compared event study approach, the comparative analysis results revealed that accounting information's predictive ability was more accurate than the machine learning algorithms’ predictive ability in stock price predictions. This suggests that using machine learning algorithms does not necessarily result in better prediction performance and that machine learning algorithms are not the replacement for accounting information in financial predictions.
 
using the empirical approach, 564 firm-year observations from 2019 to 2022 are analyzed to predict cash-holding. The researcher employed several algorithms such as decision trees, support vectors, and K-nearest neighbor, compared to multiple linear regression based on accounting information as prediction models. The empirical results showed that decision trees as complex algorithms were proven to yield higher accuracy. While, other prediction models, MLR, KN, and SV had (RMSE) and low R², which indicates the Low accuracy of predictions. Also, the empirical results reported that accounting information significantly affects the accuracy of    algorithms regarding cash-holding . Hence, the current study adds empirical evidence to related previous  review through emphasizing the complementary nature between the roles of machine learning algorithms and accounting information in financial predictions. Machine learning algorithms must be considered a supporting tool to enhance the predictive ability of accounting information, not a replacement for it. At the same time, accounting information improves the accuracy of machine learning algorithms' predictions. Therefore, it has become necessary for accountants to master  skills to maintain their jobs and assume new roles in the  of artificial .