PREDIKSI HARGA EMAS DUNIA MENGGUNAKAN METODE LONG-SHORT TERM MEMORY

Gold investment is one of the investments that quite lot interest by public and also considered safer because it has relatively low risk tends to be stable compared other instruments, especially amid uncertainty global economic conditions caused COVID-19 pandemic. Awareness about gold price predictions can provide information people who want invest in so they have higher opportunity earn profits minimize risks obtained. The prices prediction method used this study Long-Short Term Memory (LSTM) using RStudio. LSTM widely predict time series data. a variation Recurrent Neural Network (RNN) as solution overcome occurrence exploding gradient or vanishing RNN when processing long sequential best model for predicting with MAPE value 2,70601, which training data testing comparison 70% : 30% hyperparameters batch size 1, units AdaGrad optimizer, learning rate 0,1 500 epochs.