Why might a data scientist apply a logarithm to a highly right-skewed positive variable?

A. To guarantee zero missing values B. To compress large values and sometimes make relationships more regular C. To reverse the order of all observations D. To create class labels automatically