Importance of the Long-Term Seasonal Component in Day-Ahead Electricity Price Forecasting Revisited: Parameter-Rich Models Estimated via the LASSO

Recent studies suggest that decomposing a series of electricity spot prices into a trend-seasonal and a stochastic component, modeling them independently, and then combining their forecasts can yield more accurate predictions than an approach in which the same parsimonious regression or neural network-based model is calibrated to the prices themselves. Here, we show that significant accuracy gains can also be achieved in the case of parameter-rich models estimated via the least absolute shrinkage and selection operator (LASSO). Moreover, we provide insights as to the order of applying seasonal decomposition and variance stabilizing transformations before model calibration, and propose two well-performing forecast averaging schemes that are based on different approaches for modeling the long-term seasonal component.

Seasonal drivers of tuberculosis: evidence from over 100 years of notifications in Cape Town

BACKGROUND: Tuberculosis incidence varies seasonally in many settings. However, the role of seasonal variation in reactivation vs. transmission is unclear.METHODS: We reviewed data on TB notifications in Cape Town, South Africa, from 1903 to 2017 (exclusive of 1995–2002, which were unavailable). Data from 2003 onward were stratified by HIV status, age and notification status (new vs. retreatment). We performed seasonal decomposition and time-dependent spectral analysis using wavelets to assess periodicity over time. We estimated monthly peak-to-peak seasonal amplitude of notifications as a percentage of the annual notification rate.RESULTS: A seasonal trend was intermittently detected between 1904 and 1994, particularly during periods of high notification rates, but was consistently and strongly evident between 2003 and 2017, with peaks in September through November, following winter. Among young children, a second, higher seasonal peak was observed in March. Seasonal variation was greater in children (<5 years, 54%, 95% CI 47–61; 5–14 years, 63%, 95% CI 58–69) than in adults (36%, 95% CI 33–39).CONCLUSIONS: Stronger seasonal effects were seen in children, in whom progression following recent infection is known to be the predominant driver of disease. These findings may support increased transmission in the winter as an important driver of TB in Cape Town.

Deep Belief Network with Seasonal Decomposition for Solar Power Output Forecasting

Solar power is a type of renewable energy system that uses solar energy to produce electricity, and is regarded as one of the most important power sources in Taiwan. Since sunshine duration affects the amount of energy that can be generated by a solar power, the seasons of the year are important factors that should be considered for accurate solar power prediction. In the last decade, the use of artificial intelligence for forecasting systems have been quite popular, and the deep belief network (DBN) models started getting more attention. In this study, a seasonal deep belief network (SDBN) was developed to forecast monthly solar power output data. The SDBN was constructed by combining seasonal decomposition method and DBN. Further, this study used monthly solar power output data from the Taiwan Power Company. The results indicated that the proposed forecasting system demonstrated a superior performance in terms of forecasting accuracy. Also, the performance of autoregressive integrated moving average (ARIMA), generalized regression neural network (GRNN), and DBN obtained from a separate study were compared to the performance of the proposed SDBN model and showed that the latter was better than the other three models. Thus, the SDBN model can be used as an alternative method for monthly solar power output data forecasting.