Hybrid Waveform Diagnostics of Seasonal Time Series Models and Exponential Smoothing with the Application

The study aims to evaluate the effectiveness of wavelet analysis as a diagnostic tool for univariate time series models by comparing several traditional and wavelet-based methods. It also seeks to compare the performance of two different approaches for analyzing and forecasting univariate time series: traditional exponential smoothing and wavelet analysis. Additionally, the research explores the effectiveness of a hybrid model combining exponential smoothing and wavelet filters. These methods were applied to real-world data, with results showing that the hybrid model achieves higher predictive accuracy and excels in isolating noise and abrupt changes.