Long‐term intuitionistic fuzzy time series forecasting model based on vector quantisation and curve similarity measure

• DOI: 10.1049/iet-spr.2015.0496
• Publication Date: 2016-04-18T16:13:37Z
• Authors: Xiaoshi Fan; Yingjie Lei; Yanan Wang; Yanli Lu
• Keywords: Similarity measure; Centroid; Similarity (geometry)

In existing fuzzy time series forecasting models, the accuracy of excessively relies on priori knowledge and output cannot effectively forecast multi values. The reduces drastically when data deviate from experience boundary in most models. generalisation performance is insufficient. To overcome defects traditional methods, this study proposed a long-term intuitionistic (IFTS) model based vector quantisation curve similarity measure. preprocessing model, FTS theory extended to IFTS scope, raw historical are quantised vectors optimum clustering centroids searched by c -means algorithm. Curve measure algorithm procedure forecasting, which avoids influence mutation points overcomes limitation information. Euclidean distance replaced Fréchet distance, it appropriate for such directed matching. relevant models implemented three different datasets, synthetic dataset, monthly total retail sale social consumer goods daily mean temperature dataset. results, index square error average rate indicate that our better patterns than others.