A new fuzzy time series approach based on cat swarm optimization

Since there are no theoretical constraints such as the model assumption and the number of observations, the importance of the preferred fuzzy time series instead of classical time series is day by day increasing in analysis of data containing uncertainty. In obtaining predictions in fuzzy time series, basically three stages are used in the form of fuzzification, determination of fuzzy relations and defuzzification. There are many studies on the development of each of these three phases in the literature. Most of these studies are based on heuristic optimization methods such as particle swarm optimization, genetic algorithm. In this study, unlike other studies in the literature on fuzzy time series, a new approach has been proposed based on cat swarm optimization in the determination of cluster centers and support vector machine in determining of fuzzy relations. The effectiveness of this new proposed approach based on cat swarm optimization, was firstly assessed on the data commonly used in the literature. Then, respectively particle swarm optimization, genetic algorithm, artificial bee colony algorithm and ant colony algorithm were used instead of cat swarm optimization in the determination of cluster centers and these methods were applied on a new data set and the effectiveness of these methods was compared with the proposed approach. As the result of study, it has been concluded that the cat swarm optimization can be quite effective compared to many studies, especially the heuristic methods discussed in the fuzzy time series.