Gold is jewelry made from precious metals which are soft and easy to shape. The value of gold tends to fluctuate every year. Gold investments, like other investments, are not risk-free. So, investors can experience profits or even losses. Gold price predictions are needed to determine investors’ opportunities in the future. Support Vector Regression (SVR) is an application of Support Vector Machine (SVM) for regression cases whose output is real or continuous numbers. The use of the SVR method has been carried out in several studies. However, the use of the SVR method to predict the rate of gold prices in Indonesia has not been carried out. The aim of this research is to predict the rate of gold prices in Indonesia using the SVR method. The data that will be predicted is gold price data in Indonesia from October 2017 to October 2022. The SVR method is used due to the non-linear and fluctuating nature of gold price data. The study employs the Radial Basis Function (RBF) kernel considering the three parameters in the RBF kernel as gamma (𝛾), cost (C) and epsilon (𝜀)., which will be optimized using the grid search method. The results shows that the best parameters obtained are when 𝛾 = 1, C = 1, and 𝜀 = 0.1, with k = 10 because it yields the smallest error value. The level of accuracy in the prediction results is obtained using the Mean Absolute Percentage Error (MAPE) value. The MAPE value is approximately 1.28% suggesting that the prediction accuracy was very good because the MAPE value obtained was < 10%.

Haze has become a major concern as it has a significant impact over Malaysia. This haze arises by uncontrolled fires from the land clearing activity in some of the areas in asian countries that builds up during the dry season. During these periods, particulate matter (PM₁₀) was found out as the major pollutant. PM₁₀ is an air pollutant consisting of a mixture of solid and liquid particles suspended in the air where it will affect harmful to human health and environment. In this study, an analysis during the haze phenomena will be performed using PM₁₀ concentration from three areas which are Klang, Muar and Seremban in 2004, 2005, 2006 and 2009. From the study, the highest PM₁₀ concentration was observed in 2005 (Klang) with the value of 994 μg/m³ compared with others place. Then, the Multiple Linear Regression (MLR) and combination with Principle Component Analysis (PCA) and MLR are conducted. From the analysis, the PCA-MLR is better than the MLR model due to the less calculated error and better-fitted model between the predicted and the observed.

Negative Binomial Regression Analysis of Factors Influencing Stunting Cases in Central Lombok Regency