İslamoğlu, EbrucanBasaran, Murat Alper2026-01-242026-01-2420252149-3871https://search.trdizin.gov.tr/tr/yayin/detay/1306252https://doi.org/10.30783/nevsosbilen.1508663https://hdl.handle.net/20.500.12868/4239Interval Time Series (ITS) techniques are used in data analysis for modeling and forecasting. This paper introduces a hybrid model that combines two effective forecasting methods: the Modified Adaptive Network-Based Fuzzy Inference System (MANFIS) and the Elman Recurrent Neural Network (ERNN). Unlike non-interval time series data, ITS considers both the highest and lowest values within an interval to represent dynamic data, thereby capturing potential relationships between these bounds. The proposed algorithm incorporates ANFIS and ERNN structures with the following advantages: it utilizes particle swarm optimization to train the model and addresses both linear and nonlinear forecasting aspects, providing model-based and data-based approaches, respectively.The fuzzification process employs the fuzzy c-means clustering technique to systematically derive membership values from input data, thereby enhancing forecasting accuracy. The proposed method has been rigorously validated using seven diverse real-world datasets, and comparative analyses with existing algorithms in the literature confirm its superior performance.eninfo:eu-repo/semantics/openAccessParticle Swarm Optimizationİnterval-Valued Time SeriesFuzzy C-Means ClusteringERNNThe Modified ANFISArticle10.30783/nevsosbilen.15086631512232471306252