Forecasting Method for Grouped Time Series with the Use of k-Means Algorithm

TL;DR

This paper introduces a clustering-based forecasting approach for grouped time series, utilizing k-means for clustering and specialized models for centroid prediction to improve individual series forecasts.

Contribution

It proposes a novel method combining k-means clustering with centroid-based forecasting models using binary trees and a modified clonal selection algorithm.

Findings

Effective clustering of time series with k-means.

Centroid-based models improve individual series forecasting.

Demonstrated promising application for grouped time series forecasting.

Abstract

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