Author: Miguel B. Araujo; Babak Naimi
Title: Spread of SARS-CoV-2 Coronavirus likely to be constrained by climate Document date: 2020_3_16
ID: jjdtuofy_9
Snippet: Metrics of model performance on the test data were generally high (mean/SD AUC=0,76/0,03, see supplementary Figure S4 ). However, models with high performance in the test data can still generate projections that are uncertain when used for forecasting (17) . While improvements in the data and the models can reduce uncertainties, an issue we are now exploring at light of the recently published standards for applied ENMs (18) , characterizing the u.....
Document: Metrics of model performance on the test data were generally high (mean/SD AUC=0,76/0,03, see supplementary Figure S4 ). However, models with high performance in the test data can still generate projections that are uncertain when used for forecasting (17) . While improvements in the data and the models can reduce uncertainties, an issue we are now exploring at light of the recently published standards for applied ENMs (18) , characterizing the uncertainty of existing models is a first step towards understanding the limitations of projections and highlighting areas of concern. We generated 200 models by varying the initial conditions and model classes. As such, we were able to quantify and map some of the methodological uncertainties associated with projections ( Figure 5 ). Using this approach, we show that variation associated with changing initial conditions is negligible, which is unsurprising given that splits are randomly performed. In contrast, consistently with previous studies addressing climate change forecasts (19) , uncertainty associated with use of different modeling techniques is high (20, 21) . Models are considerably variable in areas projected to have low seasonal variability in climate suitability across much of Latin America, sub-Saharan Africa and South East Asia. Specific areas in India, China, Western and Central Europe, Coastal Australia, and Central USA also 6 have high levels of model variability. Such variability is probably a consequence of the sparse nature of the positive cases of COVID-19 in those areas, causing different models to adjust differently to the data. In contrast, inter-model variability across northern higher latitudes is lower and this is likely because most records between January and March are found there. Such areas of low inter-model variability coincide with areas where seasonal variation is also greater.
Search related documents:
Co phrase search for related documents- high level and model class: 1, 2
- high level and model performance: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12
- high level and modeling technique: 1, 2
- high level and positive case: 1, 2, 3, 4, 5, 6
- high level and previous study: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27
- high level and seasonal variation: 1, 2
- high level and specific area: 1, 2, 3, 4
- high performance and mean sd: 1, 2, 3, 4, 5, 6, 7, 8, 9
- high performance and model class: 1, 2, 3
- high performance and model performance: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73
- high performance and model variability: 1
- high performance and positive case: 1
- high performance and previous study: 1, 2
- high performance and seasonal variation: 1, 2
- high performance and specific area: 1, 2, 3, 4, 5
- initial condition and model class: 1, 2
- initial condition and previous study: 1
- inter model variability and model variability: 1, 2, 3, 4, 5
Co phrase search for related documents, hyperlinks ordered by date