MRSDV3 Model based on Weibull Distribution as a Model for Extreme Value

N.-J. Park, K.M. George, and N. Park (USA)


Risk analysis, risk discovery, extreme value distribution, Weibull distribution, Gumbel distribution, multiple regression


The ad hoc risk management system (ARMS) is a frame work composed of statistical model and surveyed field data for forecasting. The primary function of ARMS is dynamically forecasting or predicting a risky event (extreme value). An ’Ad-hoc risk Management System (ARMS) using Multiple Regression with Scaled Dummy Variable model’ (MRSDV1 model) and ‘Multiple Regression with Scaled Dummy Variable as a Model for Extreme Values’ (MRSDV2 model) have been recently proposed as an extended version of ARMS. These models focus on the forecast and management of risky situations (e.g., critical network security breach, a terror event and the invest timing in stock market), but they are insufficient in estimating and fitting extreme values, even though they can grasp the spike pattern from actual data. In this paper, we propose the MRSDV3 model transformation technique and its distribution as a model for analyzing extreme values. In this model, the Weibull distribution is employed to measure and enhance the efficiency and effectiveness of estimating and management of risky state (as represented by extreme values) in ad-hoc manner and the model is expected to serve as an extended theoretical foundation for ARMS.

Important Links:

Go Back