A Time-series Trend Forecast Method Based on Principal Component Analysis

作者:Wei Shudi; Zhao Huihuang
来源:3rd International Conference on Manufacturing Science and Engineering (ICMSE 2012), 2012-03-27 to 2012-03-29.
DOI:10.4028/www.scientific.net/AMR.472-475.2984

摘要

The time-series is the collection of chronological varying numerical ordered by time. It has a wide existence of image data, text data, hand-written data and the brain scan data patterns. The present research of time-series concentrates on series data transformation, similarity search, forecast, classification, clustering and Visualization etc. Furthermore the trend forecast of time-series is the major basis of other related research. This paper analyses the existing time-series forecasting methods and puts forward a new time-series method based on principal component analysis. The example tests the validity of the method of other related research.

全文