Multi-resolution Representation for Time Series Data Analysis
Our method has already been used for time series sensor data representation.
The multi-resolution piecewise linear representation algorithm based on important data points(MPLR-IDP), which segments the time series by a set of temporal feature points and maintains a high similarity between the processed subsequences and the raw data.
The efficient multi-resolution index structure storing all the selected IDPs, which can provide more accurate representation and more flexible data presentation without repeated calculation. Moreover, our proposed index structure can also be used as a useful tool for any other traditional PLR-TD and PLR-BU methods for time series multi-resolution representation.
The introduction of MPLR-IDP in detail will be given in the form of patent and paper within a month or two.
There are two simple schematic diagrams of our MPLR-IDP as follow: