Clustering the World Currency Exchange Rates Using Hierarchical Methods Based on Dynamic Time Warping
Jonatha Sousa Pimentel, Paulo Canas Rodrigues · Fluctuation and Noise Letters · 2021
The analysis of currency exchange rates is of great importance to analyze the economic health of a country. In this paper, we collect and analyze the historical data on exchange rates of all available currencies, considering the US dollar as reference. In particular, we are interested in clustering the collected daily time series by using a similarity measure based on dynamic time warping. In total, the observations of 150 currencies, between January 3, 2005 and April 30, 2020, are analyzed. The results show that the use of dynamic time warping as a distance measure results in the improvement of the interpretability of the dendrograms, when compared with standard similarity measures such as the Euclidean distance.