Ensembles of Localised Models for Time Series Forecasting

With large quantities of data typically available nowadays, forecasting models that are trained across sets of time series, known as Global Forecasting Models (GFM), are regularly outperforming traditional univariate forecasting models that work on isolated series. As GFMs usually share the same set of parameters across all time series, they often have the problem of not being localised enough to a particular series, especially in situations where datasets are heterogeneous. We study how ensembling techniques can be used with generic GFMs and univariate models to solve this issue. Our work systematises and compares relevant current approaches, namely clustering series and training separate submodels per cluster, the so-called ensemble of specialists approach, and building heterogeneous ensembles of global and local models. We fill some gaps in the approaches and generalise them to different underlying GFM model types. We then propose a new methodology of clustered ensembles where we train multiple GFMs on different clusters of series, obtained by changing the number of clusters and cluster seeds. Using Feed-forward Neural Networks, Recurrent Neural Networks, and Pooled Regression models as the underlying GFMs, in our evaluation on six publicly available datasets, the proposed models are able to achieve significantly higher accuracy than baseline GFM models and univariate forecasting methods.


page 1

page 2

page 3

page 4


Monash Time Series Forecasting Archive

Many businesses and industries nowadays rely on large quantities of time...

Improving the Accuracy of Global Forecasting Models using Time Series Data Augmentation

Forecasting models that are trained across sets of many time series, kno...

Global Models for Time Series Forecasting: A Simulation Study

In the current context of Big Data, the nature of many forecasting probl...

Ensemble Modeling for Time Series Forecasting: an Adaptive Robust Optimization Approach

Accurate time series forecasting is critical for a wide range of problem...

Multi-output Ensembles for Multi-step Forecasting

This paper studies the application of ensembles composed of multi-output...

The Effectiveness of Discretization in Forecasting: An Empirical Study on Neural Time Series Models

Time series modeling techniques based on deep learning have seen many ad...

SETAR-Tree: A Novel and Accurate Tree Algorithm for Global Time Series Forecasting

Threshold Autoregressive (TAR) models have been widely used by statistic...

Please sign up or login with your details

Forgot password? Click here to reset