I am trying to understand some paper. In there author has the sample of 140 observations with respective x and y for coordinates. He then proceeds with calculating the distance to the nearest neighbor. As the result he gets 140 distances (from which he calculates cdf). I am just not sure how he did. First of all, from what little I know of this algorithm, you need training data. Second, I am not sure how he got only 140 distances. Any suggestions? Perhaps, could somebody explain with little example. Thanks!!!