Time Series Prediction: Forecasting The Future And Understanding The Past (Santa Fe Institute Series)

The book is a summary of a time series forecasting competition that was held a number of years ago. The competition used four different kinds of time series (for example, one data set was chaotic from measurements of a laser, and another was a multidimensional physiological times series of heart beats and respiration, etc.).
The strength of the book lies in that it represents several ways to approach real time series prediction strategies in a concrete way - Invaluable, especially to researchers who may be just beginning.