Parallel construction of decision trees

Abstract

Decision trees are a well-established set of methods for classification, recognition and decision support in the machine learning, the identification, the data analysis and the situational management. The decision tree must be compact - it lessens expenses when answering questions. Moreover, compact trees have a better prognostic ability. In some applications, such as Data Mining, a dataset to be learned is very large. In those cases it is highly desirable to construct univariate decision trees within a reasonable period of time. This can be accomplished by parallelizing univariate decision tree algorithms.