While learning with data having large number of attribute, a system is easy to freeze or shut down or run for a long time. Therefore, the proposed Binary Conversion (BC) is a novel method to solve this kind of large attribute problem in machine learning. The purpose of BC is to reduce data dimensions by a binary conversion process. All the attributes are reserved but combined into few numbers of new attributes instead of that some attributes are removed. To prevent the information loss problem during the conversion, each binary type data value occupies its own digital position in BC. In addition, 4 data sets: nbuses, ACLP, MONK3, and Buseskod data are used in this study to test and compare the learning accuracies and learning time. The results indicate that the proposed BC can keep about the same level of accuracy but increase the learning efficiency.
