Solution of the Problem of Classification of Vehicles on the Basis of Statistical Estimates of Data

The method of analysis of vehicle characteristics, based on statistical estimates, is considered in the paper. A variety of vehicles are distributed according to classes and categories, so new models with new characteristics and new properties which appear every year, need to be assigned to the corresponding class. Making a decision tree is a convenient mean to make partition of vehicle models on the subsets which are associated with the corresponding classes. But a variety of classification trees can be obtained depending on the order of the selection of attributes which represent characteristics of vehicles. Solution of the problem to obtain a compact tree with the maximal purity of the class is offered. Approach to classify transport on road indicators of vehicles is used which is based on using entropy estimation and Gini index. The analytical platform Deductor was used to implement classification on the basis of described method.