Automatic classification in the case of unknown number of clusters using global density estimates
L. Schüler, Hans W. Wolff · Biometrical Journal · 1980
Abstract A clustering algorithm using a nonparametric density gradient estimate based on a trigonometric series expansion approach is investigated. The Mean‐Integrated‐Square‐Error will be evaluated, conditions for consistency and a stopping rule for the number of terms needed in the series expansion are given. An example demonstrates the utility of the method.