Density

Given one particular individual in your data base, are there many other individuals very similar to that individual? If the answer to that question is "Yes", then this individual is sitting in a high density region. On the other hand, if he is just about alone of his kind, then he is said to be sitting in a low density region. Statistics knows how to define mathematically this vague concept of density.

Why is that of any importance at all? Because if we were capable of estimating accurately the density around any point of the space of individuals, then virtually all the important problems of modelization would be solved as a consequence. Unfortunately, the known techniques of density estimation are gravely flawed, for reasons which are well undestood, but for which there is no known cure. This is the reason why commercial software rarely offers any technique of density estimation.

Let's note in passing that Neural Networks have recently made significant advances in that direction.
 

Now, in practice, your next Data Modeling project will probably ignore altogether.the issue of local density of data  But in the future, density estimation might become central in modelization.