Pendekatan Univariate Time Series Modelling untuk Prediksi Kuartalan Pertumbuhan Ekonomi Indonesia Pasca Vaksinasi COVID-19

The development of COVID-19 has had a significant negative impact on Indonesia’s economic growth based on the indicator of the value of the quarterly year of year data in 2020 and 2021. Economic growth is still experiencing a recession per first quarter with a percentage of - 2.19 percent at the beginning of 2021. The government has to take vaccination measures for the community gradually with the aim of reducing the number of sufferers of these cases. The purpose of this study is to predict economic growth quarterly after vaccination using 3 (three) univariate time series models, namely ARIMA, Holt-Winters and Dynamic Linear models for policymaking. Holt-Winters and Dynamic Linear models make it possible to handle time-series data containing trends and seasonality. The data is divided into training data and test data obtained from the ministry of finance and the Indonesian Central Statistics Agency (BPS). The goodness of the model uses MSE, MAE and U-Theil criteria. Based on the results of the analysis using the R library, the results show that the best modelling for economic growth data is the ARIMA model with the lowest MSE, MAE and U-Theil values with the difference between the models being 0.000242. The ARIMA model looks better than other models because the economic growth data only contains trends and assumes a seasonal element in the data. In addition, the Holt-Winters and Dynamic Linear models produce a forecast for Indonesia’s economic growth to still experience a recession (negative growth) in the next four quarterly data, while the ARIMA model produces a positive growth forecast in the fourth quarter.

Two-Stage Spatiotemporal Time Series Modelling Approach for Rice Yield Prediction & Advanced Agroecosystem Management

A robust forecast of rice yields is of great importance for medium-to-long-term planning and decision-making in cereal production, from regional to national level. Incorporation of spatially correlated adjacent effects in forecasting models in general, results in accurate forecast. The Space Time Autoregressive Moving Average (STARMA) is the most popular class of model in linear spatiotemporal time series modelling. However, STARMA cannot process nonlinear spatiotemporal relationships in datasets. Alternately, Time Delay Neural Network (TDNN) is a most popular machine learning algorithm to model the nonlinear pattern in data. To overcome these limitations, two-stage STARMA approach was developed to predict rice yield in some of the most intensive national rice agroecosystems in India. The Mean Absolute Percentage Errors value of proposed STARMA-II approach is lower compared to Autoregressive Moving Average (ARIMA) and STARMA model in all examined districts, while the Diebold-Mariano test confirmed that STARMA-II model is significantly different from classical approaches. The proposed STARMA-II approach is promising alternative to classical linear and nonlinear spatiotemporal time series models for estimating mixed linear and nonlinear patterns and can be advanced tool for mid-to-long-term sustainable planning and management of crop yields and patterns in agroecosystems, i.e., food supply and demand from local to regional levels.

Sentinel-2 Time Series Analysis for Identification of Underutilized Land in Europe

Biomass and bioenergy play a central role in Europe’s Green Transition. Currently, biomass is representing half of the renewable energy sources used. While the role of renewables in the energy mix is undisputed, there have been many controversial discussions on the use of biomass for energy due to the “food versus fuel” debate. Using previously underutilized lands for bioenergy is one possibility to prevent this discussion. This study supports the attempts to increase biomass for bioenergy through the provision of improved methods to identify underutilized lands in Europe. We employ advanced analysis methods based on time series modelling using Sentinel-2 (S2) data from 2017 to 2019 in order to distinguish utilized from underutilized land in twelve study areas in different bio-geographical regions (BGR) across Europe. The calculated parameters of the computed model function combined with temporal statistics were used to train a random forest classifier (RF). The achieved overall accuracies (OA) per study area vary between 80.25 and 96.76%, with confidence intervals (CI) ranging between 1.77% and 6.28% at a 95% confidence level. All in all, nearly 500,000 ha of underutilized land potentially available for agricultural bioenergy production were identified in this study, with the greatest amount mapped in Eastern Europe.

Decadal shifts in traits of reef fish communities in marine reserves

Marine reserves are known to impact the biomass, biodiversity, and functions of coral reef fish communities, but the effect of protective management on fish traits is less explored. We used a time-series modelling approach to simultaneously evaluate the abundance, biomass, and traits of eight fish families over a chronosequence spanning 44 years of protection. We constructed a multivariate functional space based on six traits known to respond to management or disturbance and affect ecosystem processes: size, diet, position in the water column, gregariousness, reef association, and length at maturity. We show that biomass increased with a log-linear trend over the time-series, but abundance only increased after 20 years of closure, and with more variation among reserves. This difference is attributed to recovery rates being dependent on body sizes. Abundance-weighted traits and the associated multivariate space of the community change is driven by increased proportions over time of the trait categories: 7–15 cm body size; planktivorous; species low in the water column; medium-large schools; and species with high levels of reef association. These findings suggest that the trait compositions emerging after the cessation of fishing are novel and dynamic.

OP321 The Scale And Variation Of The Impact Of COVID-19 On Prescribing Of Medicines In Primary Care In Wales

IntroductionPrescribing of medicines in primary care in Wales has been exceptional in 2020 due to COVID-19 and the associated changes to the delivery of health services. The changes are likely to have harmful, albeit unintended, consequences, including disruption of pharmacy stock management; unpredictable changes in prescribing; and interruption to patients’ supply of medicines and reduced medication adherence. Changes in prescribing are unlikely to be distributed evenly across the country or population. Therefore, this study aimed to identify changes in GP prescribing compared with previous years, the variation of these changes, and factors related to the variation in changes, to identify patient subgroups for whom the impact is disproportionate.MethodsWe identified medicines of interest where concerns around prescribing have been raised and, for each of these medicines, retrieved monthly prescribing data for each GP practice in Wales (N = 492). We then linked these data with other publicly available data (for example, practice size, indices of multiple deprivation, disease prevalence).We developed a novel approach to measure the impact of COVID-19 on GP prescribing. We compared observed with expected prescribing volume projected via time series modelling and differences were related to patient and practice characteristics using general estimating equations.ResultsThere was evidence of stockpiling of medicines during March 2020 (for example, oral-contraceptives and oral-anticoagulants with 11.6 and 18.5 percent increases from March 2019), followed by a short-term reduction in prescribing for oral-contraceptives (a reduction of 12.9 percent), but not oral-anticoagulants (an increase of 6.5 percent). However, GP level data show considerable deviation from the national trend for several GPs, which may be due to health and socio-demographic factors.ConclusionsCOVID-19 has had a major impact on primary care prescribing in Wales. The distribution of changes in prescribing will not be even across the country or the population. Identification of systematic variation in impacts on prescribing could identify geographical areas or patients in need of additional support to ensure uninterrupted and appropriate access to medicines.

Information technology for forecasting the financial results of insurance companies

The purpose of time series modelling is to predict future indicators based on the study and analysis of past and present data. Various time series methods are used for forecasting. The article uses econometric extrapolation research methods. Analyzed scientific works are related to extrapolation methods for forecasting time series. The dynamics of the financial formation related to results of Ukrainian insurance companies by the types of their activities have been analyzed. The main factors that determine the effectiveness are determined. It was found that the most rational approach to short-term forecasting of the financial results of insurers is the use of exponential smoothing. The optimal parameters are selected for the model of exponential smoothing of the first and second order by the method on the grid. The following indicators of the quality of the model were used: the mean value of the standard deviation of the model error to the actual data, Theils coefficient of discrepancy, mean absolute percentage error MARE. The net financial result of the activities of Ukrainian insurers was predicted, the lower and upper bounds of the forecast for 2021 for a reliability level of 0.95. To predict the net financial result of the activities of Ukrainian insurers, statistical data for 10 years from 2011 to 2020 were used, the financial results of the main (insurance and other operating) activities before tax, the results of financial activities before tax, the financial results of other ordinary activities (extraordinary events) before tax, income tax. The prototype of the software module for predicting the financial performance of insurance companies was developed in Statistica and Excel. Forecasting results based on the use of econometric modelling make it possible to identify permanent positive shifts in the domestic insurance market and the activities of insurers on it; to confirm the effectiveness of the adopted strategic and tactical financial decisions of insurance companies; to increase the efficiency of insurers management based on the results of quantitative determination the degree of influence of each factor on the formation of the financial results related to their activities; to identify trends in the development of the situation in the future, to more accurately form a set of measures to maximize profits and minimize costs of insurance companies to ensure guarantees of reliable insurance protection and satisfy the interests of their owners. Keywords: financial results; insurance companies; net financial result; exponential smoothing; time series; econometric forecasting methods.

COVIDSpread: real-time prediction of COVID-19 spread based on time-series modelling

A substantial amount of data about the COVID-19 pandemic is generated every day. Yet, data streaming, while considerably visualized, is not accompanied with modelling techniques to provide real-time insights. This study introduces a unified platform, COVIDSpread, which integrates visualization capabilities with advanced statistical methods for predicting the virus spread in the short run, using real-time data. The platform uses time series models to capture any possible non-linearity in the data. COVIDSpread enables lay users, and experts, to examine the data and develop several customized models with different restrictions such as models developed for a specific time window of the data. COVIDSpread is available here: http://vafaeelab.com/COVID19TS.html.