Andrew I want to understand what one-hot encoding means in machine learning. How is it used to convert categorical data into a format that models can process? Can someone also explain its advantages and limitations with examples?