A Causal Model with UncertainTime-Series Effect Based on Evidence Theory

Abstract

Probability and possibility are both convenient scales of uncertainty, because they are defined by a distribution function. They also have complementary properties in the sense that probability is a quantitative and objective ratio scale, while possibility is a qualitative and subjective ordinal scale. The paper discusses probabilistic and possibilistic causal models with a time-series effect from the viewpoint of Evidence theory, and shows that they can be defined by a single equation with different conditions of focal elements using the basic probability assignments. The equation could be recognized as a causal model with a general representation of uncertainty in the form of Evidence theory. The paper finalizes the discussion with the properties of the generalized uncertain causal model.