A Forecasting Method Based On K-Means Clustering and First Order Fuzzy Time Series

Abstract

Clustering is the method of partitioning or grouping a given set of designs into several clusters. Forecasting accuracy is one of the most favourable critical issues in Fuzzy Time Series models. In the past few decades, a number of forecasting models built on fuzzy time series principles have been place out. These models have been frequently used to solve many different types of problems, particularly those involving predicting issues when the event data are linguistic values. The time series forecasting is based on the historical data of 40 years. This study examines introduce a novel fuzzy time series forecasting model that takes historical data as the universe of discourse, cluster the universe of discourse using the K-means clustering approach, and then divides the clusters into intervals. K-mean clustering algorithm was applied by using the IDL software to find the centroid values. The suggested approach is used to forecast data on coal production. At the end, we compared the forecasting values of K-means clustering method and the arithmetic method.