On a distribution‐free discriminant analysis
Arjun K. Gupta, B. K. Kim · Biometrical Journal · 1978
Abstract In this paper a distribution‐free classification procedure is studied when it is assumed that the characteristic in one population is stochastically larger than in the other. The procedure utilizes the kernel type density estimators. An upper bound for the failure rate is also derived and the classification rule is shown to be consistent.