Gold Price Prediction Based on Support Vector Regression (SVR) and Artificial Neural Network (ANN) Comparison

Gold is the most popular type of investment due to its ease of ownership, low level of risk, and various advantages such as inflation-resistance, high liquidity, and affordability for all groups. Supported by a BPS survey which shows that 58% of people choose gold as the main instrument and an average global price increase trend of 12% in the last 10 years, it is important to predict gold prices to support optimal investment decision-making. This study aims to evaluate and compare the performance of the Support Vector Regression (SVR) and Artificial Neural Network (ANN) models in predicting the price of Indonesian gold.The methods used in this study are SVR and ANN methods with Grid Search Optimization carried out on each method to determine the optimal parameters. The results of this study show that the best model in predicting Indonesian gold prices is the ANN model with an activation function Tanh produced an RMSE value of 44576.672, MAPE of 2.88%, MAE of 38301.776, and AIC amounting to 752,528. The ANN method with the Tanh activation function is the best method for predicting the price of Indonesian gold with smaller RMSE, MAPE, MAE, and AIC values compared to the SVR method.