Dynamic Discretization of Continuous Values from Time Series

摘要: Two methods to assign discrete values continuous from time series, using dynamic information about the are proposed. The first method is based on a particular statistic which allows us select value for new series. second one concept of significant distance between consecutive series defined. This definition qualitative changes in values. In both methods, conversion process into opposition static classical used machine learning. Finally, we use proposed practical case. We transform daily clearness index results display that with obtained captures better sequential properties original