The rapid growth of computing technology to process vast amount of data has impelled more interest in data mining. Such interest was mainly aimed at knowledge discovery to improve decision making process in diverse range of applications, including Islamic finance. One of the most critical steps in data mining is data preprocessing, as it would directly affect the quality of insights obtained at the later stage. Feature selection has been widely used in data preprocessing phase to improve the machine-learning algorithm and model interpretability. However, there has been limited attention has been given on the evaluation of feature selection methods on its effectiveness to process input data for Induction Decision Tree (IDT). Hence, this study aims to address such gap in the literature through the use of real-world data in Islamic finance to evaluate the improvement that generated by feature selection method. The result of the study shows that the use of such technique has resulted in better performance of the IDT model generated in the study.