Pattern matching-based prediction using affine combination of two measures: Two are better than one

摘要

Time series forecasting based on pattern matching has received a lot of interest in recent years due to its simplicity and the ability to predict complex nonlinear behaviours. The choice of the metric to measure the similarity between two time series depends mainly on the specific features of the considered data and it can influence on forecasting results. In this paper, unlike the conventional method, we propose an improved pattern matching-based prediction method using a linear combination of two measures, Euclidean distance and dynamic time warping, in order to achieve a better forecasting result. These two distance measures are chosen because they are the two most commonly used metrics for pattern matching in time series. The experimental results showed that our approach can produce better results on time series forecasting work in comparison to the pattern matching-based method under Euclidean distance or dynamic time warping in terms of prediction accuracy.