The Kozachenko-Leonenko differential entropy estimator is a non-parametric estimator based on k:th nearest neighbors of a sample set.
The data must be full-dimensional. If this is not the case, consider using the Nilsson-Kleijn estimator instead.
The probability density function of the underlying random variable must be differentiable. For example, uniform distribution in gives bad estimates.
Sample Estimate of the Entropy of a Random Vector,
L. F. Kozachenko, N. N. Leonenko,
Probl. Inf. Transm., Vol. 23, No. 1, pp. 95-101,
1987.
Kozachenko-Leonenko nearest neighbor estimator
Kozachenko-Leonenko nearest neighbor estimator
Kozachenko-Leonenko nearest neighbor estimator