At WeatherSense Labs , you're developing a machine learning model to predict rainfall for a given day. Your dataset contains thousands of input features , and you want to explore ways to speed up model training while keeping the impact on accuracy minimal . What approach should you consider?