Aggregating Prophet and Seasonal Trend Decomposition for Time Series Forecasting of Italian Electricity Spot Prices

The cost of electricity and gas has a direct influence on the everyday routines of people who rely on these resources to keep their businesses running. However, the value of electricity is strongly related to spot market prices, and the arrival of winter and increased energy use owing to the demand for heating can lead to an increase in energy prices. Approaches to forecasting energy costs have been used in recent years;however, existing models are not yet robust enough due to competition, seasonal changes, and other variables. More effective modeling and forecasting approaches are required to assist investors in planning their bidding strategies and regulators in ensuring the security and stability of energy markets. In the literature, there is considerable interest in building better pricing modeling and forecasting frameworks to meet these difficulties. In this context, this work proposes combining seasonal and trend decomposition utilizing LOESS (locally estimated scatterplot smoothing) and Facebook Prophet methodologies to perform a more accurate and resilient time series analysis of Italian electricity spot prices. This can assist in enhancing projections and better understanding the variables driving the data, while also including additional information such as holidays and special events. The combination of approaches improves forecast accuracy while lowering the mean absolute percentage error (MAPE) performance metric by 18% compared to the baseline model.

Long-term health impact of PM2.5 under whole-year COVID-19 lockdown in China.

The health impact of changes in particulate matter with an aerodynamic diameter <2.5 µm (PM2.5) pollution associated with the COVID-19 lockdown has aroused great interest, but the estimation of the long-term health effects is difficult because of the lack of an annual mean air pollutant concentration under a whole-year lockdown scenario. We employed a time series decomposition method to predict the monthly PM2.5 concentrations in urban cities under permanent lockdown in 2020. The premature mortality attributable to long-term exposure to ambient PM2.5 was quantified by the risk factor model from the latest epidemiological studies. Under a whole-year lockdown scenario, annual mean PM2.5 concentrations in cites ranged from 5.4 to 68.0 µg m-3, and the national mean concentration was reduced by 32.2% compared to the 2015-2019 mean. The Global Exposure Mortality Model estimated that 837.3 (95% CI: 699.8-968.4) thousand people in Chinese cities would die prematurely from illnesses attributable to long-term exposure to ambient PM2.5. Compared to 2015-2019 mean levels, 140.2 (95% CI: 122.2-156.0) thousand premature deaths (14.4% of the annual mean deaths from 2015 to 2019) attributable to long-term exposure to PM2.5 were avoided. Because PM2.5 concentrations were still high under the whole-year lockdown scenario, the health benefit is limited, indicating that continuous emission-cutting efforts are required to reduce the health risks of air pollution. Since a similar scenario may be achieved through promotion of electric vehicles and the innovation of industrial technology in the future, the estimated long-term health impact under the whole year lockdown scenario can establish an emission-air quality-health impact linkage and provide guidance for future emission control strategies from a health protection perspective.

Robust trend estimation for COVID-19 in Brazil.

Estimating patterns of occurrence of cases and deaths related to the COVID-19 pandemic is a complex problem. The incidence of cases presents a great spatial and temporal heterogeneity, and the mechanisms of accounting for occurrences adopted by health departments induce a process of measurement error that alters the dependence structure of the process. In this work we propose methods to estimate the trend in the cases of COVID-19, controlling for the presence of measurement error. This decomposition is presented in Bayesian time series and spatio-temporal models for counting processes with latent components, and compared to the empirical analysis based on moving averages. We applied time series decompositions for the total number of deaths in Brazil and for the states of São Paulo and Amazonas, and a spatio-temporal analysis for all occurrences of deaths at the state level in Brazil, using two alternative specifications with global and regional components.