Climate variability and global warming have increasingly affected cotton production across major agricultural zones. This study presents a data-driven approach to predict cotton yield under varying climate change scenarios using the Random Forest (RF) machine learning algorithm. Historical cotton yield data were combined with climate parameters such as temperature, rainfall, humidity, and CO₂ concentration over multiple decades. The Random Forest model was trained on this multi-dimensional dataset to identify key influencing factors and generate yield predictions under future Representative Concentration Pathways (RCPs). The results demonstrate that the RF model provides robust, accurate forecasts and highlights specific climatic thresholds beyond which yields significantly decline. This model serves as a valuable decision-support tool for policymakers and farmers in climate-resilient agricultural planning.