DATA MODELING OF GOLD PRICES USING SUPPORT VECTOR REGRESSION WITH LINEAR, POLYNOMIAL, AND RADIAL BASIS FUNCTION KERNELS

Investment is a commitment of funds or other resources made with the goal of obtaining profits in the future. Gold is one of the commodities favored by investors because of its relatively stable value. Although relatively stable, gold prices experience fluctuations, which introduce risks associated with gold investments. The pattern of gold prices can be mathematically modeled using Support Vector Regression (SVR). SVR finds a function as a hyperplane (separating line) in the form of a regression function that fits all input data with an error and minimizes it as much as possible. SVR enables a balance between data fitting and overfitting through the use of kernel functions. The pattern of gold prices and SVR is the main focus of this research, aiming to model gold prices effectively. Furthermore, SVR seeks to obtain a hyperplane (regression function) that fits all input data by minimizing errors. The gold price modeling results using SVR indicated that the best model was obtained with a linear kernel, showing an error of 0.73% with parameters C = 10^(-4) and ? = 10^(-4). This implies the model's strong ability to follow data patterns accurately, resulting in highly reliable forecasting.