A machine learning engineer is analyzing a dataset and observes that the distribution of a continuous feature is heavily right-skewed. Which transformation is most likely to make the distribution approximately normal?