A Frequency-domain MLP Model for Forecasting Marine Economy Stocks under Climate Influence

This study presents FreTS, a new type of machine learning model. The aim of the system is to predict movements in the prices of ocean-related stocks. It uses something called frequency domain learning. It focuses on industries such as shipping, fisheries and marine energy. These industries are becoming more and more important in global finance and sustainable development. As climate change continues to affect these regions, it is very important to have accurate tools to forecast what will happen. Traditional models often find it hard to deal with complex data and noisy signals, especially when there is limited data available. FreTS solves this problem by applying the Discrete Fourier Transform to convert time series data into the frequency domain. This means that the model can spot patterns and trends that other models might miss. In tests using information about sea life and the climate, FreTS did better than LSTM, SARIMA and RNN models. It has a mean square error of 0.012 and a correlation coefficient of 0.958, which proves how reliable and strong it is. In today's unpredictable maritime economy, FreTS provides a practical solution for financial forecasting by dealing with noise and capturing long-term trends.