Forecasting Rupiah exchange rate by integrating PCA with neural networks.

The Covid-19 endemic is a challenge in the economic field, the endemic causes uncertainty in the exchange rate. This uncertainty causes the exchange rate to fluctuate which is characterized by data volatility. Data volatility for investors causes investors to always be ready to face the possibilities that will occur in the capital market to make decisions about their portfolios. Uncertainty in investing makes investors have to be able to predict future data accurately. To overcome this, the researchers conducted forecasting using the Neural Network (NN) method using Principal Component Analysis (PCA) as the preprocessing method. NN optimization is used with the aim of overcoming the occurrence of overfitting in the training process. The preprocessing process is used to overcome the problem of multicollinearity in the Rupiah exchange rate data used. To show the performance of using PCA and NN, MAE and MSE performance measures are used. The experimental results show that PCA+NN is able to reduce errors compared to individual NNs. [ABSTRACT FROM AUTHOR]