A new data classification method was prompted for the classify problem about samples with known prior probabilities. Vectors near the boundaries were preextracted from the training samples based on vector projection, the values of the class-conditional probability density of the boundary vectors were approximately computed by k-nearest-neighbors estimation. To approximate the class-conditional probability density function of each class of the objects in the training data set, radial basis function networks were constructed using the boundary vectors as the network centers. The classification was realized by the minimum error rate Bayesian decision rule. Simulation results for machine learning data sets show that the proposed algorithm can deliver the same level of accuracy as the support vector machines in data classification applications, and can effectively carry out data classification with more than two classes of objects.