For a random variable X which takes k values with probabilities p₁, p₂, …p_k, the entropy of X is given by $H(X)=\underset{1}{\overset{k}{{\displaystyle -\sum }}}{p}_i\text{l}\text{o}{\text{g}}_2{p}_i.$ This is the expected value of the information when sampling X. It is maximal when the probability distribution is uniform. See Shannon’s theorem.