COVID-19 epidemiological indicators POD spatial decomposition

The epidemiological indicators (prevalence, mortality, convalescence) for COVID-19 are investigated as a temporary-spatial dependencies. Proper Orthogonal Decomposition is applied for the first time to this type of data; the main modes and corresponding coefficients are obtained. Due to this method, it is shown that there are modes concentrated in the particular regions which means there are independent factors for disease spreading. Additionally, it is showed that Empirical Mode Decomposition can be successfully applied for noise-reduction and better understanding time dependencies. The exponential nature of the decomposition error decree shows the accuracy of the decomposition. Despite the ability of POD to reveal hidden dependencies, it requires rows of simultaneous data, which in fact may not be so. In the article the correction applied is discussed, however it may not be enough because of mistakes in raw data. The method is recommended to use unless the data it is applied to is inaccurate.