Dimensionality reduction for Interval Time Series

To meet some application´s requirement, we introduce a new data type, ITS (Interval Time Series), which is modified based on the widely studied date type-time series, along with a dimension reduction technique to it. The implied idea of our method is derived from the orthogonal transformation. We then present an optimization to it, with which our method can performs much better. Experiments show our method works fine for various dimensionalities and scales of dataset.