A New Pattern Representation Method for Time-Series Data

The rapid growth of Internet Things (IoT) and sensing technologies has led to an increasing interest in time-series data analysis. In many domains, detecting patterns IoT interpreting these are challenging issues. There several methods analysis that deal with issues such as volume velocity streams. However, analysing the content streams extracting insights from dynamic is still a task. this paper, we propose pattern representation method which represents frames vectors by first applying Piecewise Aggregate Approximation (PAA) then Lagrangian Multipliers. This allows representing continuous series can be used processed various higher-level methods. We introduce new change point detection uses constructed its evaluate compare our Blocks Eigenvalues Algorithm (BEATS) Symbolic approXimation (SAX) cluster datasets. have evaluated algorithm using UCR datasets also healthcare dataset. evaluation results show significant improvements proposed method.