<![CDATA[Investment is not new thing for most people, especially bitcoin which is very popular in recent years. Understanding the trend of price movements in investing is very important insight for investors to minimize investment risk, but predicting trend changes is very difficult challenge because it has fluctuating difference in value. The value of the increase and decrease in price of bitcoin is influenced by uncertainty factors such as political problems, economic problems at global level. So, need an algorithm that can predict prices in the future which is one strategy to maximize profits in investing. This study performs several processes to predict bitcoin price movements including, pre-processing, normalization, training the Long-Short Term Memory (LSTM) algorithm, evaluating regression matrix using Mean Square Error (MSE). Based on the results of tests that have been carried out in this study, LSTM algorithm can predict bitcoin price movements as evidenced by the MSE evaluation matrix value of 0.00374 with test parameters including 64 hidden_size, 18 sequence data, optimizer Adam, learning_rate of 0.005, and epoch 200. This research also involves several weight updated algorithms including Stochastic Gradient Descent (SGD), Stochastic Gradient Descent with Momentum (SGDM), and Adaptive Moment Estimation (ADAM) to find optimal prediction results.]]>
Copyrights © 2022
