Foreign direct investment (FDI) is a key economic phenomenon and a key driver of economic growth, thereby making its accurate forecasting crucial for policymakers and investors. It further brings capital, technology, and expertise into emerging markets, fostering job creation and innovation. The current study compares four machine learning models—support vector regression (SVR), a Deep Neural Network (DNN), empirical-mode-decomposition-based SVR, and an empirical-mode-decomposition-based DNN—to improve the forecasting accuracy for foreign direct investment using the exchange rate and gross domestic product as the independent variables. The empirical mode decomposition technique is applied to decomposing the series into intrinsic mode functions (IMFs) before feeding it into the machine learning model(s). The models' forecasting performance is evaluated using the mean squared error (MSE), the root mean squared error (RMSE), the mean absolute error (MAE), the mean absolute percentage error (MAPE), the symmetric mean absolute percentage error (SMAPE), and the mean bias deviation (MBD). The results demonstrate that the EMD-based SVR model outperformed all of the other models, achieving the highest accuracy due to its ability to filter noise and capture economic noise. Furthermore, it is shown that decomposition-based hybrid models are effective in financial forecasting, and they provide valuable insights for economic decision-making. Future research could explore other machine learning models and add more macroeconomic variables.