1: \begin{abstract}

2: The property of almost every point being a \leb{} point has proven to be crucial for the consistency of several classification algorithms based on nearest neighbors.

3: We characterize \leb{} points in terms of a $1$-\nn{} regression algorithm for pointwise estimation, fleshing out the role played by tie-breaking rules in the corresponding convergence problem. 

4: We then give an application of our results, proving the convergence of the risk of a large class of $1$-\nn{} classification algorithms in general metric spaces where almost every point is a \leb{} point.

5: \end{abstract}

6: