A Recent-Biased Dimension Reduction Technique for Time Series Data

There are many techniques developed for tackling time series and most of them consider every part of a sequence equally. In many applications, however, recent data can often be much more interesting and significant than old data. This paper defines new recent-biased measures for distance and energy, and proposes a recent-biased technique based on DWT for time series in which more recent data are considered more significant. With such a recent-biased technique, the dimension of time series can be reduced while effectively preserving the recent-biased energy. Our experiments have demonstrated the effectiveness of the proposed approach for handling time series.