Paper Title
Silver Price Prediction using Nonlinear Autoregression Neural Network for Thailand Silver Jewelry Industry

Abstract
This paper presents silver price prediction models for Thailand silver jewelry industry. The models are based on nonlinear autoregression neural network (NAR). The models use historical silver and gold prices as the only time-series data. Therefore, these models are suitable to be used by the manufacturers in Thailand silver jewelry industry because it only relies on the silver and gold historical prices. Specifically, we compare the accuracy of the NAR models with only silver data, and the model with gold and silver time-series data. The results show that the model with gold and silver time-series data give much better accuracy, e.g. MSE of 5.4015 × 10-5 and 4.0139 × 10-5 for silver data only and gold-and-silver data, respectively. Using both silver and gold price data is over 25.6% improvement over using only silver price. This is because the gold and silver prices are highly correlated. Consider both of them together as the time-series data provide more complete information to the NAR models.