Unclear and inconsistent findings related to temperature and humidity may be due to methodological challenges and data limitations. Similar methodological challenges were highlighted when evaluating the association between air pollution and COVID-19 outbreaks36,37. The novelty of the virus, with less than a full annual cycle of data available in most places, makes it difficult to disentangle a seasonal signal or inter-annual trends from meteorological factors using time-series models38. Moreover, different interventions (e.g. restrictions of mass gatherings, international travel and school closures) adopted by countries at different times after the onset of local outbreaks potentially confound the association between weather variables and COVID-19 spread. These challenges have led us to consider an ecological approach where we compared the outbreak curve early in the epidemic, minimising the confounding effect of NPIs. Despite this, we found a strong association of the OxCGRT Government Response Index with Re, confirming the importance of interventions implemented early on in the epidemic in controlling COVID-19 dynamics39.
Comparing the early-phase outbreak curves in different countries is challenging given that countries have different case definitions, and early COVID-19 data only captured a small portion of cases, mainly hospitalised patients or individuals with severe symptoms. The estimated high proportion of asymptomatic cases compromises the use of COVID-19 case counts to estimate transmission dynamics40. We used an estimated response variable, i.e. the effective reproduction number, calculated accounting for reporting delays and other sources of uncertainty. The 20-day duration was chosen as a compromise between needing enough days for a more precise Re estimation, while, at the same time, limiting the window to provide more constant weather, case ascertainment and Re estimates within the window. A larger window would bias estimates in ways that cannot be readily adjusted for. Our meta-analysis approach accounts for the uncertainty in Re estimates, which in turn reduces the level of certainty in the results. Further, 20 days is ~4 generations of infections, which, under most reporting scenarios, is sufficient to be confident about estimates in the level of transmission. We assume that within the 20-day time window, the case definition is constant within a city or country and Re is not affected by differences in case definition between countries.
Data in this study were obtained from a well-established MCC Collaborative Research Network50. The current MCC network covers 750 locations (cities or regions) in 43 countries/regions. For this study, 26 countries provided a daily time series of COVID-19 cases for a total of 502 locations (cities or small regions). COVID-19 data were downloaded from a publicly available repository or obtained from health agencies (Supplementary Table 1) and data management was performed using Microsoft Excel 2019. The time series from 1 January 2020 to 31 May 2020 comprises 2,771,137 COVID-19 cases, representing 44.8% of the cumulative cases registered by 31 May 2020 in the Johns Hopkins database ( ). Supplementary Table 1 shows the sources used for each country along with the definition of COVID-19 cases.
The everyday job of the actuary involves the estimation of a future series of events. Examples include the estimation of future streams of liability outgo and asset income in life assurance, the run-off of outstanding claims in nonlife insurance, and the future numbers of persons in a subgroup of the total population. This estimation can be qualitative or quantitative, short-term or long-term, deterministic or stochastic and will involve the establishment of a mathematical-statistical model, and the determination of the relevant parameters by an analysis of the data available.
Over recent years the popularity of time series has soared. Given the widespread use of modern information technology, a large number of time series may be collected during business, medical or biological operations, for example. As a consequence there has been a dramatic increase in the amount of interest in querying and mining such data, which in turn has resulted in a large number of works introducing new methodologies for indexing, classification, clustering and approximation of time series. In particular, many new distance measures between time series have been introduced. In this paper, we propose a new distance function based on a derivative. In contrast to well-known measures from the literature, our approach considers the general shape of a time series rather than point-to-point function comparison. The new distance is used in classification with the nearest neighbor rule. In order to provide a comprehensive comparison, we conducted a set of experiments, testing effectiveness on 20 time series datasets from a wide variety of application domains. Our experiments show that our method provides a higher quality of classification on most of the examined datasets.
We use R throughout the book and we intend students to learn how to forecast with R. R is free and available on almost every operating system. It is a wonderful tool for all statistical analysis, not just for forecasting. See the Using R appendix for instructions on installing and using R.
What is a time series analysis and what are the benefits? A time series analysis focuses on a series of data points ordered in time. This is one of the most widely used data science analyses and is applied in a variety of industries.
Creating a time series model in Python allows you to capture more of the complexity of the data and includes all of the data elements that might be important. It also makes it possible to make adjustments to different measurements, tuning the model to make it potentially more accurate. 2b1af7f3a8