About this chapter
Some time series contain abrupt changes or behavior at several scales that a heavily smoothed representation can obscure. Nystrom and colleagues introduce wavelet methods as a way to examine information in time and frequency together. The chapter develops wavelet theory and transforms, discusses thresholding and multiresolution analysis, and reviews forecasting approaches that use wavelets directly or as part of hybrid models. Applications include wind speed, earthquakes, and traffic, with a worked example and research questions. Readers can use it to consider what preprocessing preserves or removes before treating a cleaner-looking signal as a more useful basis for prediction.
Companion overview · About our editorial material
Carry a question into the reading
Read this contribution alongside related chapters, then work through an original companion discussion or exercise.
Companion reading noteFrom data to a defensible modelCite this contribution
Jared Nystrom; Raymond R. Hill; Andrew Geyer; Joseph J. Pignatiello Jr.; Eric Chicken. (2024). Wavelet Methods for Time Series Forecasting. In Natalie M. Scala & James P. Howard, II (Eds.), Handbook of Military and Defense Operations Research (2nd ed., pp. 345–382). Chapman & Hall/CRC. https://doi.org/10.1201/9781003396307-14
Reference details: published chapter metadata. Report a correction.
Nearby in the subject index
Other chapters in this edition sharing a companion topic.