A new predicting method based on estimate of holder exponent by continuous wavelet transform

Abstract

This project is aimed at developing a new predicting method of time series based on an estimate of the Holder exponent with the continuous wavelet transform. Analyzing time-oriented data and forecasting future values of a time series are among the most important problems at tracking in wireless sensor, face tracking, traffic flow predicting, and exchange rate fluctuation forecasting. The main proposed idea of using continuous wavelet transform is based on an estimate of singularity signal with the Holder exponent. It is observed that the time series changes in accordance with sharp changes of the Holder exponent. Results obtained with the proposed algorithm are presented and compared with state-of-the-art forecasting methods in terms of accuracy of prediction.