This R Notebook reproduces the Curse of Dimensionality¹ in the nearest neighbor regression that here is defined as the increasing of the interval size to get 10% of the data according with the increasing of dimensions. So according more dimensions are add to the domain, greater is the size of the range to get the same proportion of data points. Conforming we increase the size of the range we lost “locality” of the information, losing the capacity to resume the information with a simple average.

This post talks about the use of Mean Squared Error (MSE) against the flexibility of a function fitted as a technique to assess the model accuracy in a specific problem as describe in the An Introduction to Statistical Learning in R book.

A place to put my ideas and thoughts on the topics that I like and I get involved in everyday life, especially about software development, agility, lean production and data science.