A geometric analysis of time series leading to information encoding and a new entropy measure (Q2406318)

The authors consider a geometric analysis of time series which leads to information encoding and introduce a new entropy measure. The first section is an introductory section with almost all main results in this topic. The second section provides some theoretical background. It first starts with some information about the model and its properties which are used to understand results in the next sections. The concept of discretization of time series is discussed. Some results about the left and right product of P-operator are derived and the sign changes of P-operator are discussed. Beside this, the authors show that the 3-point neighborhood of \(s[n]\) consisting of \(n-1\), \(n\) and \(n+1\) can have 13 distinct geometric configurations. This section ends with some results about transition in time series and different configurations of continuous time series. It is proved that a continuous time series is made up of at least 29 different configurations. In the third section about applications the authors introduce a new entropy measure named the semantic entropy. At the end of the paper, the information power is introduced.