Model for reconstructing a causation process from time varying data describing an event and for predicting the evolution dynamics of the event

摘要

A method of reconstructing a causation process from time varying data describing an event, the data consisting in a certain number of entities each having a position in a space, and each of the entities being characterized by at least a quantity or value relatively to at least one feature and in the quantity or value relatively to at least one of the features of the entities at least at two different times or at each time instant of a sequence of time instants;the method describing the higher likelihood transition of all entities i, j from the time n to the time n+1 as a function of the position coordinate of the entity I and of the entity j and the quantity of the at least one feature of the entity I and of the entity j at the time n and at the time n+1:; M _i,j ^[n,n+1]=ψ(x_i, y_i, q_i^[n], q_i^[n+1], x_j, y_j, q_j^[n], q_j^[n+1]).the function determining the strength of the connection between each entity i at time n and each other entity j at time n+1;the method determining the source causing changes in quantity of the entity j from the time n to the time n+1 as the entity i for which the strength of connection is a maximum.;The invention relates also to a method of predicting the evolution dynamics of an event or process starting from the information about the causation process obtained from the above function.