Data Science

by John D. Kelleher

In fact, a theorem known as the “no free lunch theorem” (Wolpert and Macready 1997) states that there is no one best ML algorithm that on average outperforms all other algorithms across all possible data sets.