摘要

In order to extract the time-effect factors of multiple time series, a generalized time-effect factor model (GTFM) and a statistic for identifying the model are proposed. This model is a general form of the Pena-Box model. Based on the model, a pollen data set from Taiwan is analyzed. A factor, named temperature, is extracted. Its trend is then used to recover the weather of Taiwan over the past 5000 years.