Can you explain the steps you would take to preprocess a dataset for training a machine learning model?