Kernel density estimation

A nonparametric method for estimating the probability density function of a random variable from data without assuming a parametric form. A kernel function (commonly Gaussian) is centered at each observation and the results are summed and normalized; bandwidth selection controls the degree of smoothing.