Notes. Overview of the computation of the asteroid-class probability for each observation based on its cluster probabilities. The upper part of the table contains clusters whose members are mapped to a single asteroid class. The lower part contains clusters where the resulting asteroid class probabilities depend on the criterion given in the middle column. GMM(z_x, z_y) means that the cluster probability is split based on a Gaussian mixture model with N components fit to all cluster members in z_x and z_y, where N is equal to the number of possible outcome classes (i.e. each mixture component represents one candidate class). P_X(y) refers to the probability of belonging to the class or cluster X given the value of y. The last line gives the definition of the Ch-class, which is the last step of the classification.