Gradient boosting for extreme quantile regression

Extreme quantile regression provides estimates of conditional quantiles outside the range of the data. Classical quantile regression performs poorly in such cases since data in the tail region are too scarce. Extreme value theory is used for extrapolation beyond the range of observed values and estimation of conditional extreme quantiles....

Statistical post processing of extreme weather forecasts

In this thesis we develop several statistical methods to estimate high conditional quantiles to use for statistical post-processing of weather forecasts. We propose methodologies that combine theory from extreme value statistics and machine learning algorithms in order to estimate high conditional quantiles in large covariate spaces. In...

Are collision and crossing course surrogate safety indicators transferable?: A probability based approach using extreme value theory

In order to overcome the shortcomings of crash data a number of surrogate measures of safety have been developed and proposed by various researchers. One of the most widely used temporal indicators is time-to-collision (TTC) which requires the road users to be on a collision course. Road users that are strictly speaking not on a collision...

Assessing the effect of reservoir operation on seasonal flooding of the Masaka wetland, Rwanda

Fluvial flooding is a recurring phenomenon in Rwanda due to a combination of climate, topography and human interventions. This poses a hazard to human lives and infrastructure, particularly in low-lying areas adjacent to major rivers. An example of such areas is the Masaka wetland, located in the southern edge of the capital Kigali along the...

Tail dependence in financial data: Modelling dependence in dynamic factor models with copulas and extreme value theory

In this thesis we model extreme log-returns on economic variables and apply this to Ortec Finance's model. These extreme log-returns are relevant for risk management applications such as Value-at-Risk and other measures of tail risk. We use extreme value theory to simulate economic variables with the desired tail behaviour. We pay special...

Risk measures of heavy-tailed data

How disastrous would it be if a flood occurred in your country? Or a Tsunami, earthquake etc... The events above are considered to be 'extreme'. When such phenomena happen the results are usually very destructive, but the probability that such an extreme event occurs is of course very small and might not even occur in an entire lifetime. The...