A novel evolutionary approach to linear time-series forecasting model

This paper presents a handshake between the concepts of genetic algorithms and the forecasting problem to present a novel search based multi-phase genetic algorithm to the forecasting problem based on the time series model. The backbone concept of the paper lies in utilizing the genetic approach effectively for implementing the Autoregressive process, a linear stochastic model where a time series is supposed to be a linear aggregation of random shocks. We propose to utilize the concept of genetic algorithms to transform an initial population of random suggested solutions to a population that contains solutions approximating the optimal one. A carefully chosen fitness function acts in the capacity of a yardstick to appraise the quality of each "chromosome" to aid the selection phase. We simulated the presented approach on a Pentium IV processor and obtained results that were very encouraging.