The 200 billion dollar fishing industry is predicted to see a reduction of 40 to 60 billion dollars due to
climate change and the deoxygenation of the ocean. In order to understand these effects, our project aims to determine the main features in predicting oxygen loss as well as forecasting future levels of oxygen in the ocean. Using data from Copernicus Marine Service and GEBCO on the region (-100,0) longitude and (-15,35) latitude, we construct a dataset consisting of features salinity, phosphate, depth, nitrate, chlorophyll, temperature, and silicate, and we additionally include days of lag for each feature in order to introduce more dependence on recent trends. By using this dataset, we construct several models using XGBoost and Neural Networks which outperform our baseline test model. Additionally, through SHAP analysis, we determine that the most important features are salinity, phosphate, nitrate, and temperature.