Comparing the Predictive Power of Multilayer Perceptron (MLP) and Multiple Linear Regression in Estimating the Yield of Islamic Treasury Bonds

Articles in Press

Document Type : Original Article

Authors
pegah shahrokhi 1
seyyed mohammad hashemi nejad 2
Seyyed Atefeh Hoseini 3
1 Ph.D Student, Department of Accounting and Finance,Islamic Azad University,Firuzkuh Branch ,Firuzkuh,Iran
2 Assistant Professor,Department of Business Management ,Islamic Azad Univrsity,Tehran Medical Science Branch,Tehran,Iran
3 Assistant Professor,Department of Accounting,Firuzkuh Branch, Islamic Azad University, ,Firuzkuh,Iran
10.22034/jik.2025.77941.4596
Abstract
This study compares the predictive power of the Multilayer Perceptron (MLP) neural network and multiple linear regression in estimating the yield of Islamic treasury bonds. Using financial and economic data from 2018 to 2021, prediction models were designed and evaluated based on various variables affecting bond yields. The main goal was to assess the accuracy of these two methods and analyze their efficiency in financial risk management. The results showed that the MLP model outperformed multiple linear regression, offering higher accuracy with lower error in predicting bond yields. These findings indicate that neural network models, due to their ability to model complex and nonlinear relationships, serve as suitable tools for financial analysis and economic forecasting. The results from this research can enhance financial analysis and risk management of treasury bonds, and provide a foundation for developing combined approaches in this field.
Keywords
Yield estimation
MLP
Regression
treasury bonds