Statistical decisions under nonparametric a priory information

Abstract A program is developed for simular and experimental data handling. The main purposes are: the choice of the model most precisely describing the experiment, classification of particles and interaction processes. Procedures used: Bayes error calculation, K nearest neighbour density estimation, “Leave-one-out-at-a-time” test. Used nonparametric methods provide quantitative comparison of multivariate distribution mixture classification. Applications: high energy physics, cosmic ray physics.