Wavelet Methods for Time Series Forecasting

摘要: Accurate and timely time series forecasts have become increasingly important for short-term operational forecasting. However, parameter estimation and interpretation in such models has become particularly difficult due to the high volume of data produced by modern weather sensors. Wavelet methods are powerful techniques for time series denoising or feature extraction, thus facilitating greatly improved model estimation and performance. This chapter presents a survey of wavelet methods for time series analysis and forecasting, to include an examination of novel wavelet techniques from three disparate fields developed to address unique requirements.