Forecasting of time series’ groups with application of fuzzy c-mean algorithm

Abstract

Listen

Translate article

The paper is focused on the development of the forecasting method for time series’ groups with application of the clustering algorithms. Fuzzy c-means algorithm (FCM algorithm) is suggested to be a basic one for clustering. The coordinates of the clusters’ centers have been put in correspondence with summarizing time series data – the clusters’ centroids. A description of time series, the clusters’ centroids, is implemented with application of the forecasting models. They are based on the strict binary trees and the modified clonal selection algorithm. The forming possibility of analytic dependences with application of such forecasting models, is shown. It is suggested to use a common forecasting model, which is constructed for time series – the clusters’ centroids, in forecasting for the private (individual) time series in the cluster. The use advantage of FCM algorithm in comparison with k-means algorithm for the clustering of time series’ groups is shown. The promising application of the suggested forecasting method for forecasting of time series’ groups is demonstrated.