Extreme risk decision modeling: a new approach with an application to air pollution based on PM2.5
针对发生概率极低但后果严重的极端风险，提出了基于广义帕累托分布的两成分风险模型，并将其应用于欧洲某城市PM2.5浓度分析，识别出一年中风险最高的月份和时期。
Extreme risks are events or scenarios that have a very low occurrence probability, however they can imply catastrophic consequences. These risks are often beyond the range of normal expectations and can lead, depending on the context, to financial or economic losses, loss of life, or environmental damage. The main purpose is to develop extreme risk decision models that account for the presence of extreme values. Two-component risk models based on mean and standard deviation and on mean and value at risk are defined for the Generalized Pareto distribution and properties are investigated in function of the model parameters. It is also shown that the new risk models can be represented as a multi-criteria risk assessment method: the additive ratio assessment method. The risk models are applied to air pollution risk assessment by analysing fine particulate matter (PM2.5) concentrations in a European city, in order to identify the riskiest months and periods of the year.