Modeling Indonesian Interest Rates Using Trigonometric Functions: A Comparison of Heuristic and Hybrid Analytical-Heuristic Methods Andrew Nilsen (a), Udjianna Sekteria Pasaribu (b*), Adilan Widyawan Mahdiyasa (b)
a) Master Program in Actuarial Science, Faculty of Mathematics and Natural Science,
Institut Teknologi Bandung, Jl. Ganesha No. 10, Bandung 40132, Indonesia
b) Statistics Research Division, Faculty of Mathematics and Natural Science,
Institut Teknologi Bandung, Jl. Ganesha No. 10, Bandung 40132, Indonesia
*Corresponding Author: udjianna[at]math.itb.ac.id
Abstract
Modeling interest rates is important for understanding the economic environment and making informed decisions about monetary and fiscal policy. A trigonometric function can provide better predictions of future movements due to its ability to capture periodic patterns and trends. In this study, we applied both heuristic and hybrid analytical heuristic methods to estimate the parameters of such a function using Indonesian interest rate data from March 2009 to July 2016. The hybrid analytical heuristic method was least squares with basin hopping approximation, a method for finding parameters that minimizes the sum of the squared error. It is often used because it can be done analytically and is robust to the presence of outliers in the data. The heuristic method was Particle Swarm Optimization, an optimization method that is fast, efficient, easy to implement, and is used to find the optimal solution to complex problems with many variables. After estimating the parameters of the trigonometric function using both the heuristic and analytic methods, we interpolated the data to find the mean squared error MSE, mean absolute error MAE, and coefficient of determination R^2 for each method. From the results, it can be concluded that the least squares method with basin hopping approximation performed better than Particle Swarm Optimization, with an error of MSE: 4.76xE-6, MAE: 0.0017, and R^2: 0.9125, compared to the error of MSE which is 4.93xE-6, MAE: 0.0017, and R^2: 0.91 for Particle Swarm Optimization.
