When checking a feature for a linear regression model, the data shows a strongly right-skewed distribution. Which transformation is most appropriate to help satisfy the regression assumptions?

A logarithmic transformation is most appropriate for a strongly right-skewed distribution because it compresses the larger values more than the smaller values, effectively reducing the skewness and making the distribution more symmetrical (closer to normal). This helps satisfy the assumption of linearity and homoscedasticity in linear regression models. An exponential transformation would exacerbate the right skew. Polynomial transformation creates new features by raising existing features to a power, which doesn't directly address skewness in the original distribution. Sinusoidal transformation is used for cyclical data patterns, not for correcting skewness.