Using Resampling Techniques for Better Quality Discretization

摘要: Many supervised induction algorithms require discrete data, however real data often comes in both and continuous formats. Quality discretization of attributes is an important problem that has effects on accuracy, complexity, variance understandability the model. Usually, other types statistical processes are applied to subsets population as entire practically inaccessible. For this reason we argue performed a sample only estimate population. Most existing methods, partition attribute range into two or several intervals using single set cut points. In paper, introduce variants resampling technique (such bootstrap) generate candidate points thus, improving quality by providing better estimation towards Thus, goal paper observe whether type can lead points, which opens up new paradigm construction soft decision trees.