Research on Time Series Forecasting Model Based on Moore Automata

摘要

For time series data mining (TSDM), the problem of time series forecasting has attracted wide attention as solving it actually paves a way to extrapolate past behavior into the future. Researchers have long been interested in modeling the problem by linear regression, neural network, chaos, support vector machines, etc. In this paper, we explore the use of Moore automata for time series forecast modeling and demonstrate how the Moore automata can be converted to solve the problem with regression methods. The effectiveness of the proposed approach has been verified by experiments.