Welcome to the Monash, UEA & UCR
Time Series Extrinsic Regression Repository

monash uea ucr

This website aims to support research into Time Series Extrinsic Regression (TSER), a regression task of which the aim is to learn the relationship between a time series and a continuous scalar variable a task closely related to time series classification (TSC), which aims to learn the relationship between a time series and a categorical class label. We recommend you to read the paper for a detailed discussion of the datasets and their sources. If you use the results or code, please cite the paper "Chang Wei Tan, Christoph Bergmeir, Francois Petitjean, Geoffrey I. Webb, Time Series Extrinsic Regression: Predicting numeric values from time series data".

            @article{
              Tan2020TSER,
              title={Time Series Extrinsic Regression}, 
              author={Tan, Chang Wei and Bergmeir, Christoph and Petitjean, Francois and Webb, Geoffrey I},
              journal={Data Mining and Knowledge Discovery},
              pages={1--29},
              year={2021},
              publisher={Springer},
              doi={https://doi.org/10.1007/s10618-021-00745-9}
            }
          

If you just use the website, please reference the website as:

Chang Wei Tan, Christoph Bergmeir, Francois Petitjean, Daniel Schmidt, Geoffrey I. Webb, Anthony Bagnall, & Eamonn Keogh (2020). The Monash, UEA & UCR Time Series Extrinsic Regression Archive. URL http://tseregression.org/.
March 2021 Updated the LiveFuelMoistureContent dataset.
March 2021 The paper "Time Series Extrinsic Regression" is now published in Data Mining and Knowledge Discovery.
December 2020 The paper "Time Series Extrinsic Regression" is now in Press.

Datasets

The following table shows a list of time series extrinsic regression datasets. You can download the entire spreadsheet displayed below here and the whole dataset here (about 600 MB). The datasets are available in sktime .ts format. An example of loading the data can be found in our github repository.

The results page shows some baseline on these data using typical regressors.

ID Type Dataset Train Size Test Size Length Dimension Missing Values Donor/Source
1 Energy Monitoring AppliancesEnergy 96 42 144 24 No Luis Candanedo (UCI Repository)
2 Energy Monitoring HouseholdPowerConsumption1 746 694 1440 5 Yes Georges Hebrail & Alice Berard (UCI Repository)
3 Energy Monitoring HouseholdPowerConsumption2 746 694 1440 5 Yes Georges Hebrail & Alice Berard (UCI Repository)
4 Environment Monitoring BenzeneConcentration 3433 5445 240 8 Yes Saverio De Vito (UCI Repository)
5 Environment Monitoring BeijingPM25Quality 12432 5100 24 9 Yes Song Xi Chen (UCI Repository)
6 Environment Monitoring BeijingPM10Quality 12432 5100 24 9 Yes Song Xi Chen (UCI Repository)
7 Environment Monitoring LiveFuelMoistureContent 3493 1510 365 7 No LiuJun Zhu
8 Environment Monitoring FloodModeling1 471 202 266 1 No Jihane Elyahyioui
9 Environment Monitoring FloodModeling2 389 167 266 1 No Jihane Elyahyioui
10 Environment Monitoring FloodModeling3 429 184 266 1 No Jihane Elyahyioui
11 Environment Monitoring AustraliaRainfall 112186 48081 24 3 No Bureau of Meteorology Australia
12 Health Monitoring PPGDalia 43215 21482 256-512* 4 No Attila Reiss, Ina Indlekofer & Philip Schmidt (UCI Repository)
13 Health Monitoring IEEEPPG 1768 1328 1000 5 No Zhilin Zhang (IEEE Signal Processing Cup 2015)
14 Health Monitoring BIDMCRR 5471 2399 4000 2 No Peter Charlton & Marco Pimentel (PhysioNet)
15 Health Monitoring BIDMCHR 5550 2399 4000 2 No Peter Charlton & Marco Pimentel (PhysioNet)
16 Health Monitoring BIDMCSpO2 5550 2399 4000 2 No Peter Charlton & Marco Pimentel (PhysioNet)
17 Sentiment Analysis NewsHeadlineSentiment 58213 24951 144 3 No Nuno Moniz & Luís Torgo (UCI Repository)
18 Sentiment Analysis NewsTitleSentiment 58213 24951 144 3 No Nuno Moniz & Luís Torgo (UCI Repository)
19 Forecasting Covid3Month 140 61 84 1 No covid19.who.int

* These datasets have equal length series but differs between dimensions.

Results

cd
Critical difference diagram for 13 algorithms. Solid bar means that there is no significant difference in rank between methods (see Demsar 2006). Tests are performed with the two-tailed Nemenyi test.

The following is the raw result for each of the dataset and regressor. The same results can be obtained from here

Dataset Name FPCR FPCR-Bspline SVR Optimised RandomForest XGBoost 1-NN-ED 5-NN-ED 1-NN-DTWD 5-NN-DTWD Rocket FCN ResNet InceptionNetwork
AppliancesEnergy 5.405052 5.405052 3.454574 3.4551198 3.489024 5.231953 4.227438 6.036547 4.019873 2.2990312 2.865684 3.065047 4.43533
AustraliaRainfall 8.436335 8.436336 8.650856 8.389541 8.492986 30.254139 10.232841 12.001981 11.95073 8.124137333 8.425874 8.179173 8.841251
BeijingPM10Quality 99.725946 99.732125 110.574226 94.072344 93.138127 139.22979 115.669411 139.134908 115.502744 120.0577646 94.348729 95.489374 96.749997
BeijingPM25Quality 69.379217 69.369892 71.437076 63.301428 59.495865 88.193545 74.156382 88.256082 72.717689 62.769655 59.726847 64.462746 62.227924
BenzeneConcentration 11.088396 11.094974 4.790901 0.855559 0.6377256 6.535685 5.84498 4.983578 4.868465 3.360614 4.988295 4.0612608 1.584852
BIDMC32HR 13.980558 13.980597 13.39297 15.016468 13.963799 14.836506 14.756088 15.29101 15.127008 13.9443828 13.130665 10.74142 9.424679
BIDMC32RR 3.364777 3.364704 3.17366 4.350314 4.367828 4.387345 4.134685 3.529111 3.432247 4.0929006 3.577775 3.921214 3.018405
BIDMC32SpO2 4.953519 4.953517 4.796855 4.570262 4.450805 5.530202 5.407875 5.215027 5.123964 5.221737 5.968337 5.987832 5.57612
Covid3Month 0.044912 0.044912 0.06584 0.0424 0.044682 0.05306 0.041815 0.052735 0.042943 0.0438782 0.07434 0.095338 0.053769
FloodModeling1 0.018853 0.018853 0.046304 0.015891 0.0159712 0.01482 0.016193 0.011689 0.009801 0.002356 0.006709 0.008868 0.01743
FloodModeling2 0.019079 0.019079 0.075804 0.014095 0.018199 0.018552 0.018586 0.016356 0.016238 0.005881 0.006719 0.013939 0.00729
FloodModeling3 0.021458 0.021458 0.035032 0.020429 0.0207038 0.019947 0.020765 0.01375 0.013337 0.004064 0.007873 0.01558 0.00821
HouseholdPowerConsumption1 147.548998 147.5492 152.391358 248.858964 231.089829 473.932736 432.594707 427.04311 297.221675 132.798779 162.244492 193.207281 153.716402
HouseholdPowerConsumption2 46.925185 46.929783 55.98083 46.932139 44.3729326 71.479369 64.272956 58.802634 51.494969 32.607104 46.829256 39.080121 39.409826
IEEEPPG 31.381214 31.381212 37.254146 32.10907 31.487901 33.208862 27.111213 37.140393 33.572786 36.5154892 34.325728 33.150985 23.903929
LFMC 37.683857 37.688074 39.733527 32.1626252 32.441886 47.836798 38.535526 39.971707 35.185301 29.4097538 33.25722 30.3516564 28.796294
NewsHeadlineSentiment 0.142273 0.142272 0.142917 0.147582 0.142486 0.202821 0.156636 0.197937 0.155839 0.142244 0.148065 0.150024 0.150014
NewsTitleSentiment 0.138126 0.138126 0.138881 0.143103 0.138336 0.193318 0.15095 0.187257 0.150564 0.138059 0.138082 0.138295 0.158558
PPGDalia 20.674488 20.674486 19.005216 17.530628 16.58273 21.876567 18.282277 26.024576 20.768389 14.050544 13.038805 11.382165 9.923701

Publications

  1. C. Tan, C. Bergmeir, F. Petitjean, and G. Webb, "Time Series Extrinsic Regression: Predicting numeric values from time series data" in Data Mining and Knowledge Discovery 2021 doi
  2. C. Tan, C. Bergmeir, F. Petitjean, and G. Webb, "Monash University, UEA, UCR Time Series Extrinsic Regression Archive"

About Us

We are a group of time series researchers from Monash University, University of East Anglia and University of California Riverside:

changwei
Chang Wei Tan
tony
Anthony Bagnall
christoph
Christoph Bergmeir
daniel
Daniel Schmidt
eamonn
Eamonn Keogh
francois
François Petitjean
geoff
Geoff Webb