Monash Time Series Forecasting Archive

Abstract

Many businesses and industries nowadays rely on large quantities of time series data making time series forecasting an important research area. Global forecasting models that are trained across sets of time series have shown a huge potential in providing accurate forecasts compared with the traditional univariate forecasting models that work on isolated series. However, there are currently no comprehensive time series archives for forecasting that contain datasets of time series from similar sources available for the research community to evaluate the performance of new global forecasting algorithms over a wide variety of datasets. In this paper, we present such a comprehensive time series forecasting archive containing 20 publicly available time series datasets from varied domains, with different characteristics in terms of frequency, series lengths, and inclusion of missing values. We also characterise the datasets, and identify similarities and differences among them, by conducting a feature analysis. Furthermore, we present the performance of a set of standard baseline forecasting methods over all datasets across eight error metrics, for the benefit of researchers using the archive to benchmark their forecasting algorithms.

Multi-Objective Model Selection for Time Series Forecasting

“…Datasets are taken from various forecasting competitions, the UCI (Dua & Graff, 2017), and the Monash time series forecasting repository (Godahewa et al, 2021). Dataset sources, descriptions, basic statistics, and an explanation of the data preparation procedure can be found in Appendix B.…”

“…Our benchmark provides 44 datasets which were obtained from multiple sources. 16 datasets were obtained though GluonTS (Alexandrov et al, 2020), 24 datasets are taken from the Monash Time Series Forecasting Repository (Godahewa et al, 2021), and the remaining 4 datasets were originally published as part of Kaggle 2 forecasting competitions. Table 3 provides basic statistics about all datasets.…”

“…Australian Electricity Demand (O' Hara-Wild et al, 2021;Godahewa et al, 2021) provides the 30-minute electricity demand of five Australian states (New South Wales, Queensland, South Australia, Tasmania, Victoria). Time series range from January 2002 to April 2015.…”

“…Bitcoin (Godahewa et al, 2021) provides over a dozen of potential influencers of the price of Bitcoin. Among others, these influencers are daily hash rate, block size, or mining difficulty.…”

“…CIF 2016 (Štěpnička & Burda, 2017;Godahewa et al, 2021) is the dataset from the "Computational Intelligence in Forecasting" competition in 2016. One third of the time series originate from the banking sector while the remaining two thirds are generated artificially.…”