Kernel Density Estimation

One of the main methodological approaches to hotspot analysis is kernel density estimation. Kernel density estimation builds an estimated density using sample data and two parameters known as the kernel function and the bandwidth value. The estimated density is, like any distribution, essentially a guideline for the behavior of a random variable. Here, we mean how frequently the random variable takes on any specific value, . When dealing with hotspot analysis where the data is typically geographic, the estimated density answers the question How frequently do specific longitude and latitude pairs appear?. If a specific longitude and latitude pair, , and other nearby pairs occur with high frequency, then the estimated density built using the sample data will be expected to show that the area around the longitude and latitude pair has a high likelihood.

Kernel density estimation is referred to as a smoothing algorithm, because the process of estimating a density is...